LLM API pricing, tracked from launch.
486 language models across 97 providers — input, output, and cached rates per 1M tokens, updated daily, with full price history.
Price table
| Model | Input /1M | Output /1M | Cached in /1M | Context | Released | ||
|---|---|---|---|---|---|---|---|
| $150.00 | $600.00 | — | 200K | Mar 2025 | |||
| $75.00 | $300.00 | — | 200K | Mar 2025 | |||
| $30.00 | $180.00 | — | 1.05M | Mar 2026 | |||
| $30.00 | $180.00 | — | 1M | Apr 2026 | |||
| $21.00 | $168.00 | — | 400K | Dec 2025 | |||
| $15.00 | $120.00 | — | 400K | Oct 2025 | |||
| $15.00 | $90.00 | — | 1.05M | Mar 2026 | |||
| $15.00 | $90.00 | — | 1.05M | Apr 2026 | |||
| $10.50 | $84.00 | — | 400K | Dec 2025 | |||
| $20.00 | $80.00 | — | 200K | Jun 2025 | |||
| $20.00 | $80.00 | — | 200K | Jun 2025 | |||
| $7.50 | $60.00 | — | 400K | Oct 2025 | |||
| $10.00 | $50.00 | $1.00 | 1.05M | Sep 2026 | |||
| $10.00 | $50.00 | $0.25 | 1M | Sep 2026 | |||
| $10.00 | $50.00 | $1.00 | 1M | Jun 2026 | |||
| $10.00 | $40.00 | — | 200K | Jun 2025 | |||
| $10.00 | $40.00 | $2.50 | 200K | Oct 2025 | |||
| $7.50 | $37.50 | $0.75 | 200K | Aug 2025 | |||
|
S
Fugu Ultra v2 is a multi-agent orchestration model that routes requests across specialized language models rather than relying on a single monolithic architecture.
$5.00/$30.00I/O
1M ctx
Sep 2026
|
$5.00 | $30.00 | $0.5 | 1M | Sep 2026 | ||
| $7.50 | $30.00 | $3.75 | 200K | Dec 2024 | |||
|
S
Fugu Ultra is a general-purpose language model from Sakana with a 1M-token context window.
$5.00/$30.00I/O
1M ctx
Jun 2026
|
$5.00 | $30.00 | $0.5 | 1M | Jun 2026 | ||
| $10.00 | $30.00 | — | 128K | Jan 2024 | |||
| $5.00 | $30.00 | $0.5 | 1M | Apr 2026 | |||
| $5.00 | $25.00 | $0.5 | 1.05M | Sep 2026 | |||
| $5.00 | $25.00 | $0.5 | 1.05M | Sep 2026 | |||
| $5.00 | $25.00 | $0.125 | 1M | Sep 2026 | |||
| $5.00 | $25.00 | $0.5 | 1M | Jun 2026 | |||
| $5.00 | $25.00 | $0.5 | 1M | Jul 2026 | |||
| $5.00 | $25.00 | $0.5 | 200K | Nov 2025 | |||
| $5.00 | $25.00 | $0.5 | 1M | Feb 2026 | |||
| $5.00 | $25.00 | $0.5 | 1M | Apr 2026 | |||
| $5.00 | $25.00 | $0.5 | 1M | May 2026 | |||
| $4.00 | $20.00 | $0.2 | 1.05M | Oct 2026 | |||
| $4.00 | $20.00 | $0.2 | 1M | Sep 2026 | |||
| $0.169 | $20.00 | $0.159 | 1.05M | Jul 2026 | |||
| $4.00 | $20.00 | $0.4 | 1.05M | Jul 2026 | |||
|
F
Ember-1 is a reasoning model from Fireworks Research built on Kimi K3, designed to produce efficient reasoning traces with a 1M token context window.
$3.00/$15.00I/O
1.05M ctx
Sep 2026
|
$3.00 | $15.00 | $0.3 | 1.05M | Sep 2026 | ||
| $5.00 | $15.00 | — | 128K | Apr 2024 | |||
| $2.50 | $15.00 | $0.25 | 1.05M | Apr 2026 | |||
| $5.00 | $15.00 | — | 128K | May 2024 | |||
|
P
Perplexity's general-purpose LLM with a 200k token context window, positioned as the standard tier in the Sonar line.
$3.00/$15.00I/O
200K ctx
Mar 2025
|
$3.00 | $15.00 | — | 200K | Mar 2025 | ||
| $2.50 | $15.00 | $0.25 | 1.05M | Mar 2026 | |||
| $3.00 | $15.00 | $0.3 | 1M | Sep 2025 | |||
| $3.00 | $15.00 | $0.3 | 1M | Feb 2026 | |||
| $1.75 | $14.00 | $0.175 | 400K | Dec 2025 | |||
| $1.75 | $14.00 | $0.175 | 400K | Jan 2026 | |||
| $1.75 | $14.00 | $0.175 | 400K | Feb 2026 | |||
| $1.75 | $14.00 | $0.175 | 128K | Mar 2026 | |||
| $2.50 | $12.50 | $0.25 | 200K | Nov 2025 | |||
| $2.50 | $12.50 | $0.25 | 1M | Feb 2026 | |||
| $2.50 | $12.50 | $0.25 | 1M | Apr 2026 | |||
| $2.50 | $12.50 | $0.25 | 1M | May 2026 | |||
| $2.50 | $12.50 | $0.25 | 1M | Jul 2026 | |||
|
A
Nova Premier 1.0 is Amazon's flagship general-purpose language model with a 1-million-token context window.
$2.50/$12.50I/O
1M ctx
Oct 2025
|
$2.50 | $12.50 | $0.625 | 1M | Oct 2025 | ||
| $4.00 | $12.00 | $0.5 | 1M | Sep 2026 | |||
| $2.00 | $12.00 | $0.2 | 1.05M | Jul 2026 | |||
| $2.00 | $12.00 | $0.2 | 1M | Nov 2025 | |||
| $2.00 | $12.00 | $0.2 | 1M | Feb 2026 | |||
| $2.28 | $11.40 | $0.228 | 1.05M | Jul 2026 | |||
| $2.00 | $10.00 | $0.2 | 1.05M | Oct 2026 | |||
| $2.00 | $10.00 | $0.1 | 1.05M | Oct 2026 | |||
| $2.00 | $10.00 | $0.1 | 1.05M | Sep 2026 | |||
| $2.00 | $10.00 | $0.1 | 1M | Sep 2026 | |||
| $2.00 | $10.00 | $0.1 | 1M | Sep 2026 | |||
| $2.00 | $10.00 | $0.2 | 1.05M | Sep 2026 | |||
| $2.00 | $10.00 | $0.2 | 1.05M | Jul 2026 | |||
| $2.00 | $10.00 | $0.2 | 1.05M | Jul 2026 | |||
| $2.00 | $10.00 | $0.2 | 1M | Jun 2026 | |||
| $2.50 | $10.00 | — | 128K | Apr 2024 | |||
| $2.50 | $10.00 | $1.25 | 128K | Aug 2024 | |||
| $2.50 | $10.00 | — | 128K | Aug 2024 | |||
|
I
Inflection 3 Pi is a general-purpose language model from Inflection with an 8000-token context window.
$2.50/$10.00I/O
8K ctx
Oct 2024
|
$2.50 | $10.00 | — | 8K | Oct 2024 | ||
|
I
Inflection 3 Productivity is a general-purpose LLM with an 8000-token context window, designed for a range of text generation and analysis tasks.
$2.50/$10.00I/O
8K ctx
Oct 2024
|
$2.50 | $10.00 | — | 8K | Oct 2024 | ||
| $2.50 | $10.00 | $1.25 | 128K | Nov 2024 | |||
| $2.50 | $10.00 | — | 128K | Mar 2025 | |||
| $2.50 | $10.00 | — | 256K | Mar 2025 | |||
| $1.25 | $10.00 | $0.125 | 128K | Aug 2025 | |||
| $1.25 | $10.00 | $0.125 | 400K | Sep 2025 | |||
| $1.25 | $10.00 | $0.13 | 400K | Nov 2025 | |||
| $1.25 | $10.00 | $0.125 | 400K | Nov 2025 | |||
| $1.25 | $10.00 | $0.125 | 400K | Dec 2025 | |||
| $2.50 | $10.00 | — | 128K | Jan 2026 | |||
| $1.25 | $10.00 | $0.125 | 1M | Jun 2025 | |||
| $1.25 | $10.00 | $0.125 | 1M | Aug 2025 | |||
| $1.50 | $9.00 | $0.15 | 1M | May 2026 | |||
|
Z
GLM 5.3 Prime is Z.ai's high-throughput variant of GLM 5.3, delivering 1.5–2× faster output generation while retaining full model capabilities and 1M-token context.
$2.80/$8.80I/O
1M ctx
Sep 2026
|
$2.80 | $8.80 | $0.56 | 1M | Sep 2026 | ||
|
X
MiMo-V2.6-Pro-UltraSpeed is Xiaomi's optimized inference variant of the MiMo-V2.6-Pro foundation model, tuned for faster throughput while maintaining output quality.
$4.35/$8.70I/O
1.05M ctx
Sep 2026
|
$4.35 | $8.70 | $0.036 | 1.05M | Sep 2026 | ||
|
A
Aion-1.0 is a large language model from AionLabs with a 131k-token context window, released in early 2025 before the company's shift toward multi-model systems.
$4.00/$8.00I/O
131K ctx
Feb 2025
|
$4.00 | $8.00 | — | 131K | Feb 2025 | ||
|
P
Perplexity's research-focused variant in the Sonar line, equipped with extended reasoning and search capabilities to handle in-depth research tasks.
$2.00/$8.00I/O
128K ctx
Mar 2025
|
$2.00 | $8.00 | — | 128K | Mar 2025 | ||
|
P
Perplexity's reasoning-focused variant within the Sonar line, equipped with extended chain-of-thought capabilities and a 128k token context window.
$2.00/$8.00I/O
128K ctx
Mar 2025
|
$2.00 | $8.00 | — | 128K | Mar 2025 | ||
|
A
Jamba Large 1.7 is a hybrid architecture model from AI21 that combines Transformers with Mamba state-space layers to handle extended context efficiently.
$2.00/$8.00I/O
256K ctx
Aug 2025
|
$2.00 | $8.00 | — | 256K | Aug 2025 | ||
| $2.00 | $8.00 | $0.5 | 200K | Oct 2025 | |||
| $2.00 | $8.00 | $0.5 | 1M | Apr 2025 | |||
| $2.00 | $8.00 | $0.5 | 200K | Apr 2025 | |||
|
U
Pareto is a multimodal composite model that handles text and images alongside general-purpose reasoning tasks.
$2.50/$7.50I/O
262K ctx
Sep 2026
|
$2.50 | $7.50 | $0.25 | 262K | Sep 2026 | ||
| $1.50 | $7.50 | $0.15 | 1M | Feb 2026 | |||
| $1.50 | $7.50 | $0.15 | 1M | Sep 2025 | |||
| $1.25 | $7.50 | $0.125 | 1.05M | Mar 2026 | |||
| $2.50 | $7.50 | — | 1M | May 2026 | |||
| $1.50 | $7.50 | — | 128K | Apr 2026 | |||
| $0.875 | $7.00 | $0.0875 | 400K | Dec 2025 | |||
|
Z
GLM 5.2 is Z.ai's foundation model with a 1M-token context window for processing extended documents and complex reasoning tasks.
$0.0341/$7.00I/O
1.05M ctx
Jun 2026
|
$0.0341 | $7.00 | $0.033 | 1.05M | Jun 2026 | ||
| $1.03 | $6.16 | — | 262K | Apr 2026 | |||
|
A
A multi-model roleplaying and storytelling system from AionLabs built on the GLM family of models, using collaborative generation from specialized models to handle complex narrative tasks.
$3.00/$6.00I/O
262K ctx
Sep 2026
|
$3.00 | $6.00 | $0.75 | 262K | Sep 2026 | ||
| $2.00 | $6.00 | $0.5 | 500K | Sep 2026 | |||
|
S
Fugu Max is a multi-agent orchestration model that routes requests across specialized language models to balance performance and efficiency.
$2.00/$6.00I/O
1M ctx
Sep 2026
|
$2.00 | $6.00 | $0.25 | 1M | Sep 2026 | ||
| $2.00 | $6.00 | $0.25 | 1.01M | Aug 2026 | |||
| $1.00 | $6.00 | $0.1 | 1.05M | Jul 2026 | |||
| $2.00 | $6.00 | $0.5 | 500K | Aug 2026 | |||
| $2.00 | $6.00 | $0.25 | 1.01M | Aug 2026 | |||
| $1.00 | $6.00 | $0.1 | 1.05M | Jul 2026 | |||
| $2.00 | $6.00 | $0.25 | 1M | Aug 2026 | |||
| $1.00 | $6.00 | — | 1.05M | Feb 2026 | |||
| $2.00 | $6.00 | $0.3 | 500K | Jul 2026 | |||
|
A
Aion-3.0 is a multi-model system from AionLabs that combines specialized models for collaborative generation, building on the reasoning and coding strengths of Aion-2.0 with expanded capability across language tasks.
$3.00/$6.00I/O
131K ctx
Jul 2026
|
$3.00 | $6.00 | $0.75 | 131K | Jul 2026 | ||
| $2.00 | $6.00 | $0.2 | 65K | Apr 2024 | |||
| $2.00 | $6.00 | $0.2 | 131K | Nov 2024 | |||
|
W
Palmyra X5 is Writer's large language model with a 1 million token context window.
$0.6/$6.00I/O
1.04M ctx
Jan 2026
|
$0.6 | $6.00 | — | 1.04M | Jan 2026 | ||
| $1.00 | $5.00 | $0.05 | 1.05M | Oct 2026 | |||
| $1.00 | $5.00 | $0.05 | 1.05M | Oct 2026 | |||
| $1.00 | $5.00 | $0.05 | 1.05M | Sep 2026 | |||
| $1.00 | $5.00 | $0.05 | 1.05M | Sep 2026 | |||
| $1.00 | $5.00 | $0.05 | 1M | Sep 2026 | |||
| $1.00 | $5.00 | $0.1 | 1.05M | Sep 2026 | |||
| $1.00 | $5.00 | $0.1 | 1.05M | Sep 2026 | |||
| $1.25 | $5.00 | $0.625 | 128K | May 2024 | |||
| $0.625 | $5.00 | $0.0625 | 400K | Sep 2025 | |||
| $0.625 | $5.00 | $0.125 | 1.05M | Jun 2025 | |||
| $0.625 | $5.00 | $0.0625 | 400K | Aug 2025 | |||
| $0.625 | $5.00 | $0.0625 | 400K | Nov 2025 | |||
| $1.00 | $5.00 | $0.1 | 1M | Jun 2026 | |||
|
A
Magnum v4 72B is a 72-billion-parameter language model from Anthracite Org with a 32K token context window.
$2.50/$5.00I/O
32K ctx
Oct 2024
|
$2.50 | $5.00 | — | 32K | Oct 2024 | ||
| $1.00 | $5.00 | $0.1 | 200K | Oct 2025 | |||
| $0.75 | $4.50 | $0.075 | 1.05M | May 2026 | |||
| $0.75 | $4.50 | $0.075 | 400K | Mar 2026 | |||
| $1.10 | $4.40 | $0.55 | 200K | Feb 2025 | |||
| $1.10 | $4.40 | $0.275 | 200K | Apr 2025 | |||
| $1.10 | $4.40 | $0.275 | 200K | Apr 2025 | |||
| $1.25 | $4.25 | $0.15 | 1.05M | Oct 2026 | |||
| $1.25 | $4.25 | $0.15 | 1.05M | Sep 2026 | |||
| $1.25 | $4.25 | $0.15 | 1.05M | Aug 2026 | |||
| $1.25 | $4.25 | — | 1M | Jul 2026 | |||
| $1.36 | $4.18 | $0.14 | 1.05M | Oct 2026 | |||
| $0.825 | $4.12 | $0.0825 | 1.05M | Sep 2026 | |||
|
T
Open-weight multimodal mixture-of-experts model with 41B active parameters from 975B total and a 524K token context window.
$1.00/$4.05I/O
524K ctx
Jul 2026
|
$1.00 | $4.05 | $0.17 | 524K | Jul 2026 | ||
|
T
An open-weight multimodal mixture-of-experts model with 12B active parameters from 276B total parameters and a 524K token context window.
$1.00/$4.05I/O
524K ctx
Jul 2026
|
$1.00 | $4.05 | $0.17 | 524K | Jul 2026 | ||
|
S
Sakana Namazu is a general-purpose language model with a 262k-token context window.
$0.95/$4.00I/O
262K ctx
Aug 2026
|
$0.95 | $4.00 | $0.15 | 262K | Aug 2026 | ||
| $1.00 | $4.00 | $0.25 | 1.05M | Apr 2025 | |||
| $1.00 | $4.00 | $0.25 | 200K | Apr 2025 | |||
| $0.95 | $4.00 | $0.19 | 262K | Jun 2026 | |||
|
Z
GLM 5V Turbo is Z.ai's optimized inference variant for multimodal processing with text, image, and video inputs.
$1.20/$4.00I/O
202K ctx
Apr 2026
|
$1.20 | $4.00 | $0.24 | 202K | Apr 2026 | ||
| $3.00 | $4.00 | — | 16K | Aug 2023 | |||
| $0.4 | $4.00 | — | 131K | Sep 2025 | |||
|
Z
GLM 5 Turbo is Z.ai's optimized inference variant released in February 2026, designed for faster execution with a 202K-token context window.
$1.20/$4.00I/O
202K ctx
Mar 2026
|
$1.20 | $4.00 | $0.24 | 202K | Mar 2026 | ||
| $0.95 | $4.00 | $0.16 | 256K | Apr 2026 | |||
| $0.78 | $3.90 | — | 262K | Feb 2026 | |||
| $0.75 | $3.75 | — | 262K | Apr 2026 | |||
| $0.75 | $3.75 | $0.075 | 1.05M | Aug 2026 | |||
| $0.75 | $3.75 | $0.075 | 1.05M | Jul 2026 | |||
|
N
A 55-billion-parameter mixture-of-experts model from NVIDIA with a hybrid Transformer-Mamba architecture and 512K token context window.
$0.6/$3.60I/O
512K ctx
Jun 2026
|
$0.6 | $3.60 | $0.2 | 512K | Jun 2026 | ||
| $0.5 | $3.60 | $0.3 | 262K | Feb 2026 | |||
| $0.45 | $3.50 | — | 128K | Jul 2025 | |||
|
S
A routing model that directs requests to other LLMs based on task characteristics and cost-performance tradeoffs.
$0.85/$3.40I/O
131K ctx
Jul 2025
|
$0.85 | $3.40 | — | 131K | Jul 2025 | ||
| $0.65 | $3.25 | $0.13 | 1M | Sep 2025 | |||
|
U
Pareto is a multimodal composite model built for research, coding, and agentic workflows, while delivering frontier-level performance across a broad range of general-purpose tasks.
$0.8/$3.20I/O
1.05M ctx
Oct 2026
|
$0.8 | $3.20 | $0.03 | 1.05M | Oct 2026 | ||
|
A
Nova Pro 1.0 is Amazon's general-purpose language model with a 300K-token context window, positioned between the smaller Nova Lite and the more capable Nova Premier 1.0.
$0.8/$3.20I/O
300K ctx
Dec 2024
|
$0.8 | $3.20 | — | 300K | Dec 2024 | ||
|
Z
GLM 5 is Z.ai's foundation model with a 198K-token context window, released in early 2026 as an intermediate step before the expanded-context GLM 5.1 and subsequent releases.
$1.00/$3.20I/O
202K ctx
Feb 2026
|
$1.00 | $3.20 | $0.2 | 202K | Feb 2026 | ||
| $0.32 | $3.20 | — | 262K | Apr 2026 | |||
| $0.6725 | $3.10 | $0.15 | 262K | Jun 2026 | |||
|
Z
GLM 5.1 is Z.ai's foundation model with a 200K-token context window for extended document processing and multi-turn reasoning.
$0.966/$3.04I/O
200K ctx
Apr 2026
|
$0.966 | $3.04 | $0.1794 | 200K | Apr 2026 | ||
|
E
Glimpse is a multimodal visual reasoning model from Elorian, built for visual understanding and complex visual reasoning in the physical world. It reasons directly over images and video rather than...
$0.6/$3.00I/O
262K ctx
Oct 2026
|
$0.6 | $3.00 | $0.06 | 262K | Oct 2026 | ||
|
B
Seed-2.0-Code is a specialized coding model from ByteDance Seed with a 262k token context window, designed for software development tasks.
$0.5/$3.00I/O
262K ctx
Jul 2026
|
$0.5 | $3.00 | — | 262K | Jul 2026 | ||
|
S
A 70B parameter model from Sao10K based on Llama 3.1, designed for creative and roleplay tasks with a 16K token context window.
$3.00/$3.00I/O
16K ctx
Jan 2025
|
$3.00 | $3.00 | — | 16K | Jan 2025 | ||
|
N
Hermes 4 405B is a 405-billion parameter instruction-tuned language model from Nous with a 131K token context window.
$1.00/$3.00I/O
131K ctx
Aug 2025
|
$1.00 | $3.00 | — | 131K | Aug 2025 | ||
|
R
Relace Search is a specialized model designed to locate and extract relevant code segments and documentation from large codebases and repositories.
$1.00/$3.00I/O
256K ctx
Dec 2025
|
$1.00 | $3.00 | — | 256K | Dec 2025 | ||
| $0.6 | $3.00 | $0.1 | 256K | Jan 2026 | |||
| $0.5 | $3.00 | $0.05 | 1M | Dec 2025 | |||
|
K
A coding-focused language model from Kwaipilot with a 256K token context window.
$0.74/$2.96I/O
262K ctx
Jul 2026
|
$0.74 | $2.96 | $0.15 | 262K | Jul 2026 | ||
|
S
Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters). It performs strongly in software engineering and professional...
$1.00/$2.70I/O
1M ctx
Oct 2026
|
$1.00 | $2.70 | $0.1 | 1M | Oct 2026 | ||
| $0.26 | $2.60 | — | 262K | Feb 2026 | |||
|
T
Hy4 preview is a mixture-of-experts model from Tencent with 49B active parameters and a 1M-token context window, designed for coding, tool use, and multi-step reasoning tasks.
$0.834/$2.50I/O
1.05M ctx
Aug 2026
|
$0.834 | $2.50 | $0.042 | 1.05M | Aug 2026 | ||
| $0.5 | $2.50 | $0.025 | 1.05M | Oct 2026 | |||
| $0.5 | $2.50 | $0.025 | 1.05M | Oct 2026 | |||
|
M
Moondream 3.1 is a vision-language model from Moondream, the successor to Moondream 3 Preview, with 9B total parameters and 2B active (mixture-of-experts). It handles visual question answering, captioning, OCR, and...
$0.3/$2.50I/O
32K ctx
Oct 2026
|
$0.3 | $2.50 | — | 32K | Oct 2026 | ||
|
M
Moondream 3 Preview is a vision-language model from Moondream with 9B total parameters and 2B active (mixture-of-experts). It handles visual question answering, captioning, OCR, and visual reasoning over images, with...
$0.3/$2.50I/O
32K ctx
Oct 2026
|
$0.3 | $2.50 | — | 32K | Oct 2026 | ||
| $0.2 | $2.50 | $0.09 | 262K | Aug 2026 | |||
|
B
Seed 2.1 Turbo is a multimodal model from ByteDance Seed designed for coding tasks and agentic workflows, with a 262k token context window.
$0.5/$2.50I/O
262K ctx
Aug 2026
|
$0.5 | $2.50 | — | 262K | Aug 2026 | ||
| $0.5 | $2.50 | $0.05 | 200K | Oct 2025 | |||
| $0.3 | $2.50 | $0.03 | 1.05M | Jul 2026 | |||
| $0.7 | $2.50 | — | 64K | Jan 2025 | |||
| $0.6 | $2.50 | — | 262K | Sep 2025 | |||
| $0.6 | $2.50 | $0.15 | 262K | Nov 2025 | |||
|
A
Nova 2 Lite is Amazon's lightweight multimodal model that processes text, images, and video with a 1-million-token context window.
$0.3/$2.50I/O
1M ctx
Dec 2025
|
$0.3 | $2.50 | — | 1M | Dec 2025 | ||
| $1.25 | $2.50 | $0.2 | 2M | Mar 2026 | |||
| $1.25 | $2.50 | $0.2 | 2M | Mar 2026 | |||
| $1.25 | $2.50 | $0.2 | 1M | Apr 2026 | |||
| $0.3 | $2.50 | $0.03 | 1M | Jun 2025 | |||
| $0.2 | $2.40 | — | 81K | Aug 2025 | |||
| $0.6 | $2.40 | — | 128K | Jan 2026 | |||
|
N
A 49-billion-parameter model from NVIDIA optimized for reasoning and agentic workflows, with a 262K token context window.
$0.3/$2.40I/O
262K ctx
Mar 2026
|
$0.3 | $2.40 | $0.15 | 262K | Mar 2026 | ||
|
N
A 55-billion-parameter mixture-of-experts model from NVIDIA with a hybrid Transformer-Mamba architecture and 512K token context window.
$0.6/$2.40I/O
202K ctx
Jun 2026
|
$0.6 | $2.40 | $0.12 | 202K | Jun 2026 | ||
| $0.57 | $2.30 | — | 131K | Jul 2025 | |||
| $0.375 | $2.25 | $0.0375 | 400K | Mar 2026 | |||
| $0.55 | $2.20 | $0.275 | 200K | Jan 2025 | |||
| $0.55 | $2.20 | $0.275 | 200K | Feb 2025 | |||
| $0.55 | $2.20 | $0.1375 | 200K | Apr 2025 | |||
| $0.55 | $2.20 | $0.1375 | 200K | Apr 2025 | |||
|
Z
GLM 5.2 (batch) is Z.ai's reasoning model optimized for asynchronous batch processing with a 1M-token context window.
$0.7/$2.20I/O
1.05M ctx
Jun 2026
|
$0.7 | $2.20 | $0.07 | 1.05M | Jun 2026 | ||
|
M
MiniMax M1 is a large language model with a 1 million token context window from MiniMax's earlier model lineups.
$0.55/$2.20I/O
1M ctx
Jun 2025
|
$0.55 | $2.20 | — | 1M | Jun 2025 | ||
|
Z
GLM 4.5 is Z.ai's foundation model released in mid-2025 with a 131K-token context window.
$0.6/$2.20I/O
131K ctx
Jul 2025
|
$0.6 | $2.20 | $0.11 | 131K | Jul 2025 | ||
|
Z
GLM 4.6 is Z.ai's foundation model with a 198K-token context window, released in late 2025 as an incremental update to the GLM 4.5 series.
$0.55/$2.20I/O
204K ctx
Sep 2025
|
$0.55 | $2.20 | $0.11 | 204K | Sep 2025 | ||
| $0.5 | $2.15 | $0.35 | 163K | May 2025 | |||
| $0.18 | $2.10 | — | 131K | Oct 2025 | |||
| $0.26 | $2.08 | — | 262K | Feb 2026 | |||
|
Z
GLM 5.3 (batch) is Z.ai's reasoning model optimized for asynchronous batch processing with a 1M-token context window.
$0.45/$2.00I/O
1.05M ctx
Aug 2026
|
$0.45 | $2.00 | $0.1 | 1.05M | Aug 2026 | ||
| $1.00 | $2.00 | $0.16 | 1M | Apr 2026 | |||
| $1.50 | $2.00 | — | 4K | Sep 2023 | |||
| $1.00 | $2.00 | — | 4K | Jan 2024 | |||
| $0.4 | $2.00 | $0.04 | 131K | May 2025 | |||
| $0.25 | $2.00 | $0.025 | 400K | Aug 2025 | |||
| $0.4 | $2.00 | $0.04 | 131K | Aug 2025 | |||
| $0.25 | $2.00 | $0.03 | 400K | Nov 2025 | |||
| $0.4 | $2.00 | $0.04 | 262K | Dec 2025 | |||
|
B
A general-purpose model from ByteDance Seed with 262k token context window, superseded by the Seed-2.0 family released in 2026.
$0.25/$2.00I/O
262K ctx
Dec 2025
|
$0.25 | $2.00 | — | 262K | Dec 2025 | ||
|
B
A lightweight general-purpose model from ByteDance Seed with a 262k token context window, positioned as a faster inference alternative within the Seed-2.0 family.
$0.25/$2.00I/O
262K ctx
Mar 2026
|
$0.25 | $2.00 | — | 262K | Mar 2026 | ||
|
X
MiMo-V2.5 is Xiaomi's general-purpose language model with a 1M token context window for processing extended text inputs.
$0.4/$2.00I/O
1M ctx
Apr 2026
|
$0.4 | $2.00 | $0.08 | 1M | Apr 2026 | ||
| $1.00 | $2.00 | $0.2 | 256K | May 2026 | |||
| $0.66 | $1.98 | $0.022 | 1.05M | Aug 2026 | |||
|
Z
GLM 4.7 is Z.ai's foundation model with a 202K-token context window released in late 2025.
$0.54/$1.98I/O
204K ctx
Dec 2025
|
$0.54 | $1.98 | $0.099 | 204K | Dec 2025 | ||
| $0.325 | $1.95 | — | 1M | Apr 2026 | |||
| $0.9553 | $1.91 | $0.0796 | 1.02M | Apr 2026 | |||
|
M
Morph V3 Large is a code-focused model with a 262k token context window from Morph.
$0.9/$1.90I/O
262K ctx
Jul 2025
|
$0.9 | $1.90 | — | 262K | Jul 2025 | ||
| $0.21 | $1.90 | $0.1 | 131K | Sep 2025 | |||
| $0.375 | $1.88 | $0.0375 | 1.05M | Sep 2026 | |||
| $0.375 | $1.88 | $0.0375 | 1.05M | Aug 2026 | |||
| $0.375 | $1.88 | $0.0375 | 1.05M | Jul 2026 | |||
| $0.455 | $1.82 | — | 131K | Apr 2025 | |||
|
Z
GLM 4.5V is Z.ai's multimodal variant supporting text, image, and video inputs with a 65K-token context window.
$0.6/$1.80I/O
65K ctx
Aug 2025
|
$0.6 | $1.80 | $0.11 | 65K | Aug 2025 | ||
|
X
MiMo-V2.5-Pro is Xiaomi's professional-tier language model with a 1M token context window for processing extended text inputs.
$0.48/$1.80I/O
262K ctx
Apr 2026
|
$0.48 | $1.80 | $0.096 | 262K | Apr 2026 | ||
| $0.3 | $1.80 | — | 1M | Apr 2026 | |||
|
A
Aion-RP 1.0 is an 8-billion-parameter language model from AionLabs with a 32k-token context window, released before the company's shift to multi-model systems.
$0.8/$1.60I/O
32K ctx
Feb 2025
|
$0.8 | $1.60 | — | 32K | Feb 2025 | ||
|
A
Aion-2.0 is a large language model from AionLabs with a 131k-token context window, released between the company's initial Aion-1.0 models and its shift to multi-model systems starting with Aion-3.0.
$0.8/$1.60I/O
131K ctx
Feb 2026
|
$0.8 | $1.60 | $0.2 | 131K | Feb 2026 | ||
| $0.4 | $1.60 | $0.1 | 1M | Apr 2025 | |||
| $0.26 | $1.56 | — | 1M | Feb 2026 | |||
|
P
Perceptron Mk1.5 is an embodied reasoning model that processes text, image, video, and audio to produce text responses with structured spatial annotations like points, boxes, polygons, and tracks.
$0.15/$1.50I/O
36K ctx
Sep 2026
|
$0.15 | $1.50 | — | 36K | Sep 2026 | ||
| $0.3 | $1.50 | $0.15 | 192K | Sep 2026 | |||
| $0.25 | $1.50 | — | 1.05M | Dec 2025 | |||
| $0.5 | $1.50 | — | 16K | May 2023 | |||
| $0.25 | $1.50 | $0.025 | 1.05M | May 2026 | |||
|
P
A 32k-token context model from Perceptron designed for general-purpose language tasks.
$0.15/$1.50I/O
32K ctx
May 2026
|
$0.15 | $1.50 | — | 32K | May 2026 | ||
| $0.5 | $1.50 | — | 262K | Dec 2025 | |||
| $0.36 | $1.43 | — | 256K | Sep 2025 | |||
|
A
Aion 3.5 Mini is a multi-model roleplaying and storytelling system from AionLabs built on the GLM family of models, offering a 262k-token context window in a smaller, lower-cost form.
$0.7/$1.40I/O
262K ctx
Sep 2026
|
$0.7 | $1.40 | $0.18 | 262K | Sep 2026 | ||
|
Z
GLM 5.3 is Z.ai's full-size foundation model with a 1M-token context window, released in August 2026 as the latest general-purpose reasoning and generation model.
$0.0699/$1.40I/O
1.05M ctx
Aug 2026
|
$0.0699 | $1.40 | $0.0689 | 1.05M | Aug 2026 | ||
|
A
Aion-3.0-Mini is a smaller parameter variant of AionLabs' multi-model system that combines specialized models for collaborative generation, designed around roleplaying and storytelling tasks.
$0.7/$1.40I/O
131K ctx
Jul 2026
|
$0.7 | $1.40 | $0.18 | 131K | Jul 2026 | ||
|
A
Aion-1.0-Mini is a smaller parameter variant of AionLabs' Aion-1.0 with a 131k-token context window, released in early 2025.
$0.7/$1.40I/O
131K ctx
Feb 2025
|
$0.7 | $1.40 | — | 131K | Feb 2025 | ||
| $0.32 | $1.28 | $0.064 | 1M | Jun 2026 | |||
|
Z
GLM 5.3 FlashX is Z.ai's high-speed variant of GLM 5.3 Flash, a native multimodal model built on hybrid sparse and linear attention architecture.
$0.37/$1.25I/O
1.05M ctx
Sep 2026
|
$0.37 | $1.25 | $0.09 | 1.05M | Sep 2026 | ||
| $0.15 | $1.25 | $0.03 | 1.05M | Jun 2025 | |||
| $0.15 | $1.25 | $0.015 | 1.05M | Jul 2026 | |||
|
B
ERNIE 4.5 VL 424B A47B is Baidu's multimodal model that processes both text and images with a 123,000-token context window.
$0.42/$1.25I/O
123K ctx
Jun 2025
|
$0.42 | $1.25 | — | 123K | Jun 2025 | ||
|
R
Relace Apply 3 is a general-purpose LLM with a 256k token context window for processing large documents and code repositories.
$0.85/$1.25I/O
256K ctx
Sep 2025
|
$0.85 | $1.25 | — | 256K | Sep 2025 | ||
|
D
Cogito v2.1 is a 671-billion parameter language model with a 128k token context window.
$1.25/$1.25I/O
128K ctx
Nov 2025
|
$1.25 | $1.25 | — | 128K | Nov 2025 | ||
| $0.2 | $1.25 | $0.02 | 400K | Mar 2026 | |||
|
T
Open-weight multimodal mixture-of-experts model with 12B active parameters out of 276B total and a 524K token context window, optimized for batch processing.
$0.5/$1.20I/O
524K ctx
Jul 2026
|
$0.5 | $1.20 | $0.1 | 524K | Jul 2026 | ||
| $0.3 | $1.20 | $0.04 | 131K | Aug 2026 | |||
|
T
Open-weight multimodal mixture-of-experts model with 12B active parameters from 276B total and 524K token context.
$0.45/$1.20I/O
524K ctx
Jul 2026
|
$0.45 | $1.20 | $0.1 | 524K | Jul 2026 | ||
|
M
MiniMax M3 is a multimodal foundation model that processes text, image, and video inputs to produce text output, with a 524K token context window.
$0.3/$1.20I/O
524K ctx
May 2026
|
$0.3 | $1.20 | $0.06 | 524K | May 2026 | ||
|
M
A sparse mixture-of-experts model from Meituan with 48B active parameters designed to handle very long contexts up to 1M tokens.
$0.3/$1.20I/O
1.05M ctx
Jul 2026
|
$0.3 | $1.20 | $0.006 | 1.05M | Jul 2026 | ||
| $0.2 | $1.20 | $0.02 | 1.05M | Jul 2026 | |||
|
A
Virtuoso Large is a general-purpose language model from Arcee AI with a 131k token context window.
$0.75/$1.20I/O
131K ctx
May 2025
|
$0.75 | $1.20 | — | 131K | May 2025 | ||
|
M
Morph V3 Fast is a smaller, faster variant of the Morph V3 line with an 81k token context window.
$0.8/$1.20I/O
262K ctx
Jul 2025
|
$0.8 | $1.20 | — | 262K | Jul 2025 | ||
| $0.15 | $1.20 | — | 262K | Sep 2025 | |||
|
M
MiniMax M2.1 is a large language model with a 204K token context window, released as part of MiniMax's M2 generation lineup.
$0.3/$1.20I/O
204K ctx
Dec 2025
|
$0.3 | $1.20 | $0.03 | 204K | Dec 2025 | ||
|
M
MiniMax M2-her is a large language model with a 65K token context window, positioned as a lightweight tier within MiniMax's M2 generation.
$0.3/$1.20I/O
65K ctx
Jan 2026
|
$0.3 | $1.20 | $0.03 | 65K | Jan 2026 | ||
|
M
MiniMax M2.5 is a large language model with a 204K token context window, positioned as a general-purpose productivity model within MiniMax's lineup.
$0.3/$1.20I/O
204K ctx
Feb 2026
|
$0.3 | $1.20 | $0.03 | 204K | Feb 2026 | ||
|
M
MiniMax M2.7 is a large language model positioned between the M2.5 productivity model and the M3 multimodal generation, with a 204K token context window.
$0.3/$1.20I/O
204K ctx
Mar 2026
|
$0.3 | $1.20 | $0.06 | 204K | Mar 2026 | ||
|
K
A code-focused language model from Kwaipilot with a 256K token context window.
$0.3/$1.20I/O
262K ctx
Mar 2026
|
$0.3 | $1.20 | $0.06 | 262K | Mar 2026 | ||
|
M
MiniMax M3 is a multimodal foundation model that processes text, image, and video inputs to produce text output, with a 524K token context window.
$0.3/$1.20I/O
524K ctx
May 2026
|
$0.3 | $1.20 | $0.06 | 524K | May 2026 | ||
|
S
Step 3.7 Flash is a sparse Mixture of Experts model from StepFun that activates a subset of its parameters per token, designed for fast inference across a 256K token context.
$0.2/$1.15I/O
256K ctx
May 2026
|
$0.2 | $1.15 | $0.04 | 256K | May 2026 | ||
| $0.29 | $1.14 | $0.11 | 128K | Mar 2025 | |||
| $0.1875 | $1.12 | — | 1M | Apr 2026 | |||
|
M
MiniMax-01 is a large language model with a 1 million token context window, positioned as MiniMax's longest-context text model predecessor to the multimodal M3 generation.
$0.2/$1.10I/O
1M ctx
Jan 2025
|
$0.2 | $1.10 | — | 1M | Jan 2025 | ||
| $0.09 | $1.10 | — | 262K | Sep 2025 | |||
|
P
INTELLECT-3 is a large language model from Prime Intellect with a 131k token context window.
$0.2/$1.10I/O
131K ctx
Nov 2025
|
$0.2 | $1.10 | — | 131K | Nov 2025 | ||
|
M
MiniMax M2 is a large language model with a 204K token context window from MiniMax's M2 generation.
$0.255/$1.02I/O
204K ctx
Oct 2025
|
$0.255 | $1.02 | — | 204K | Oct 2025 | ||
|
R
Beam 1.0 is Reflection's open-weight reasoning model for coding and long-horizon agentic workflows. It uses a sparse mixture-of-experts architecture with 501B total and 23B active parameters per token, and a...
$0.25/$1.00I/O
1.05M ctx
Oct 2026
|
$0.25 | $1.00 | $0.05 | 1.05M | Oct 2026 | ||
| $0.2 | $1.00 | $0.02 | 131K | Aug 2025 | |||
| $0.125 | $1.00 | $0.0125 | 400K | Aug 2025 | |||
|
N
Nex-N2-Pro is a large language model from Nex AGI with a 262K token context window, positioned as an earlier generation before the 2026 N2.5 agentic release.
$0.25/$1.00I/O
262K ctx
Jun 2026
|
$0.25 | $1.00 | $0.025 | 262K | Jun 2026 | ||
|
N
A 405-billion parameter instruction-tuned language model from Nous with a 131K token context window.
$1.00/$1.00I/O
131K ctx
Aug 2024
|
$1.00 | $1.00 | — | 131K | Aug 2024 | ||
| $0.66 | $1.00 | — | 32K | Nov 2024 | |||
|
P
Perplexity's general-purpose LLM with a 127k token context window, positioned as the standard tier in the Sonar line.
$1.00/$1.00I/O
127K ctx
Jan 2025
|
$1.00 | $1.00 | — | 127K | Jan 2025 | ||
| $0.8 | $1.00 | $0.4 | 128K | Feb 2025 | |||
| $0.3 | $1.00 | $0.1 | 262K | Jul 2025 | |||
| $0.27 | $1.00 | — | 163K | Sep 2025 | |||
| $0.29 | $1.00 | — | 262K | Oct 2025 | |||
| $0.14 | $1.00 | $0.05 | 262K | Feb 2026 | |||
| $0.195 | $0.975 | $0.039 | 1M | Sep 2025 | |||
| $0.39 | $0.97 | — | 262K | Apr 2026 | |||
| $0.25 | $0.95 | $0.13 | 163K | Aug 2025 | |||
| $0.1 | $0.95 | $0.05 | 262K | Apr 2026 | |||
| $0.23 | $0.92 | — | 262K | Jun 2026 | |||
|
V
An instruction-tuned 24B model based on Mistral that minimizes refusals on sensitive or controversial topics.
$0.2/$0.9I/O
128K ctx
Jul 2025
|
$0.2 | $0.9 | — | 128K | Jul 2025 | ||
| $0.3 | $0.9 | $0.03 | 256K | Aug 2025 | |||
|
Z
GLM 4.6V is Z.ai's multimodal model supporting text, image, and video inputs with a 131K-token context window.
$0.3/$0.9I/O
131K ctx
Dec 2025
|
$0.3 | $0.9 | $0.055 | 131K | Dec 2025 | ||
|
X
Xiaomi's flagship foundation model built at over 1 trillion parameters, supporting a 1M token context window for extended text processing.
$0.435/$0.87I/O
1.05M ctx
Sep 2026
|
$0.435 | $0.87 | $0.0036 | 1.05M | Sep 2026 | ||
| $0.435 | $0.87 | $0.0036 | 1M | Apr 2026 | |||
|
S
A 70-parameter model from Sao10K based on Llama 3.1, tuned for creative and roleplay tasks with a 131K token context window.
$0.85/$0.85I/O
131K ctx
Aug 2024
|
$0.85 | $0.85 | — | 131K | Aug 2024 | ||
|
Z
GLM 4.5 Air is Z.ai's lightweight variant of GLM 4.5 with a 131K-token context window, designed for faster inference at a reduced capability tier.
$0.13/$0.85I/O
131K ctx
Jul 2025
|
$0.13 | $0.85 | $0.025 | 131K | Jul 2025 | ||
| $0.2 | $0.8 | $0.05 | 1.05M | Apr 2025 | |||
| $0.8 | $0.8 | — | 8K | Jan 2025 | |||
|
T
Skyfall 36B V2 is a 36-billion-parameter model from TheDrummer with a 32K token context window.
$0.55/$0.8I/O
32K ctx
Mar 2025
|
$0.55 | $0.8 | $0.25 | 32K | Mar 2025 | ||
|
A
Coder Large is a code-focused language model from Arcee AI with a 32k token context window.
$0.5/$0.8I/O
32K ctx
May 2025
|
$0.5 | $0.8 | — | 32K | May 2025 | ||
| $0.12 | $0.8 | $0.07 | 262K | Feb 2026 | |||
|
A
A large reasoning model from Arcee AI with a 262k token context window, designed to handle extended reasoning tasks and complex problem-solving.
$0.25/$0.8I/O
262K ctx
Apr 2026
|
$0.25 | $0.8 | $0.06 | 262K | Apr 2026 | ||
| $0.26 | $0.78 | $0.052 | 1M | Feb 2025 | |||
| $0.26 | $0.78 | — | 1M | Sep 2025 | |||
|
I
Mercury 2.5 Preview is an early version of Inception's diffusion-based language model that generates and refines multiple tokens in parallel.
$0.2/$0.75I/O
260K ctx
Aug 2026
|
$0.2 | $0.75 | $0.02 | 260K | Aug 2026 | ||
| $0.175 | $0.75 | $0.02 | 131K | Aug 2026 | |||
| $0.25 | $0.75 | $0.025 | 262K | Dec 2025 | |||
| $0.25 | $0.75 | — | 16K | May 2023 | |||
| $0.125 | $0.75 | $0.0125 | 1.05M | May 2026 | |||
|
M
Weaver is an alpha-stage language model from Mancer with an 8,000-token context window.
$0.4/$0.75I/O
8K ctx
Aug 2023
|
$0.4 | $0.75 | — | 8K | Aug 2023 | ||
|
S
A 70-parameter model based on Llama 3.3, created by Sao10K for creative and roleplay tasks with a 131K token context window.
$0.65/$0.75I/O
131K ctx
Dec 2024
|
$0.65 | $0.75 | — | 131K | Dec 2024 | ||
| $0.25 | $0.75 | $0.12 | 262K | Oct 2025 | |||
|
I
Mercury 2 is a large language model from Inception with a 128k token context window.
$0.25/$0.75I/O
128K ctx
Mar 2026
|
$0.25 | $0.75 | $0.025 | 128K | Mar 2026 | ||
| $0.51 | $0.74 | — | 8K | Apr 2024 | |||
|
L
LFM2.5-24B-A2B is a mixture-of-experts reasoning model from Liquid AI with 24B total and 2.45B active parameters. It always reasons before answering and is built for agents that call tools. Liquid...
$0.05/$0.7I/O
128K ctx
Oct 2026
|
$0.05 | $0.7 | $0.025 | 128K | Oct 2026 | ||
|
N
A 70-billion parameter instruction-tuned language model from Nous with a 131K token context window.
$0.7/$0.7I/O
131K ctx
Aug 2024
|
$0.7 | $0.7 | — | 131K | Aug 2024 | ||
|
U
A 13B parameter model created by merging multiple base models using spherical linear interpolation (SLERP), designed to balance instruction-following with general capability.
$0.35/$0.65I/O
6K ctx
Jul 2023
|
$0.35 | $0.65 | — | 6K | Jul 2023 | ||
| $0.65 | $0.65 | — | 8K | Jul 2024 | |||
| $0.2156 | $0.6468 | $0.0069 | 1.05M | Aug 2026 | |||
| $0.1 | $0.625 | $0.01 | 400K | Mar 2026 | |||
|
I
Ling-2.6-1T is a trillion-parameter instruct model from inclusionAI with a 262K-token context window.
$0.075/$0.625I/O
262K ctx
Apr 2026
|
$0.075 | $0.625 | $0.015 | 262K | Apr 2026 | ||
|
I
Ring-2.6-1T is a trillion-parameter instruct model from inclusionAI with a 262K-token context window.
$0.075/$0.625I/O
262K ctx
May 2026
|
$0.075 | $0.625 | $0.015 | 262K | May 2026 | ||
|
M
A mixture-of-experts model with 8 expert pathways and 22 billion parameters per expert, developed by Microsoft for handling complex reasoning and generation tasks.
$0.62/$0.62I/O
65K ctx
Apr 2024
|
$0.62 | $0.62 | — | 65K | Apr 2024 | ||
|
M
MiniMax-M3.1-Flash is a multimodal model from MiniMax. Compared with M3, it delivers significant improvements across multiple benchmarks, particularly in complex agentic tasks, coding, and cybersecurity. It is built for complex,...
$0.15/$0.6I/O
1.05M ctx
Oct 2026
|
$0.15 | $0.6 | $0.015 | 1.05M | Oct 2026 | ||
| $0.15 | $0.6 | $0.003 | 1.05M | Oct 2026 | |||
| $0.1 | $0.6 | $0.01 | 1.05M | Jul 2026 | |||
|
U
Solar Pro 4 is a large language model from Upstage with a 524K token context window.
$0.15/$0.6I/O
524K ctx
Aug 2026
|
$0.15 | $0.6 | $0.03 | 524K | Aug 2026 | ||
| $0.1 | $0.6 | $0.01 | 1.05M | Jul 2026 | |||
|
K
A code-focused language model from Kwaipilot with a 256k token context window.
$0.15/$0.6I/O
256K ctx
Jul 2026
|
$0.15 | $0.6 | $0.03 | 256K | Jul 2026 | ||
| $0.15 | $0.6 | — | 128K | Mar 2024 | |||
| $0.15 | $0.6 | $0.075 | 128K | Jul 2024 | |||
| $0.2 | $0.6 | $0.02 | 32K | Feb 2025 | |||
| $0.15 | $0.6 | — | 128K | Mar 2025 | |||
| $0.15 | $0.6 | $0.075 | 131K | Aug 2025 | |||
| $0.15 | $0.6 | — | 262K | Oct 2025 | |||
|
U
A large language model from Upstage with a 131K token context window.
$0.15/$0.6I/O
131K ctx
Jan 2026
|
$0.15 | $0.6 | $0.015 | 131K | Jan 2026 | ||
|
T
Hy3 preview is Tencent's mixture-of-experts reasoning model with a 262k-token context window.
$0.18/$0.6I/O
262K ctx
Apr 2026
|
$0.18 | $0.6 | $0.06 | 262K | Apr 2026 | ||
| $0.15 | $0.6 | — | 1M | Apr 2025 | |||
|
T
Hunyuan A13B Instruct is a 13-billion-parameter instruction-tuned language model from Tencent with a 131k-token context window.
$0.14/$0.57I/O
131K ctx
Jul 2025
|
$0.14 | $0.57 | — | 131K | Jul 2025 | ||
| $0.351 | $0.555 | — | 128K | Mar 2025 | |||
| $0.09 | $0.55 | — | 262K | Jul 2025 | |||
|
T
Hy3 is Tencent's mixture-of-experts reasoning model with a 262k-token context window.
$0.132/$0.528I/O
262K ctx
Jul 2026
|
$0.132 | $0.528 | $0.033 | 262K | Jul 2026 | ||
|
L
LFM2.5-2.6B is a compact reasoning model from Liquid AI. It is suited for agent workflows, data extraction, RAG, and long-context processing. Liquid advises against using it for agentic coding or...
$0.35/$0.5I/O
128K ctx
Aug 2026
|
$0.35 | $0.5 | $0.04 | 128K | Aug 2026 | ||
|
N
NVIDIA Nemotron 3.5 Super VL is an open multimodal mixture-of-experts model from NVIDIA, with 12B active parameters out of 120B total. It takes text, images, documents, GUI screenshots, and video...
$0.15/$0.5I/O
262K ctx
Oct 2026
|
$0.15 | $0.5 | $0.02 | 262K | Oct 2026 | ||
| $0.1 | $0.5 | $0.01 | 1M | Oct 2026 | |||
| $0.1 | $0.5 | $0.01 | 1.05M | Sep 2026 | |||
|
P
A 27B-parameter model from PrismML based on Qwen, compressed with ternary quantization to reduce size while maintaining reasoning capability across coding, mathematics, and tool use.
$0.075/$0.5I/O
262K ctx
Sep 2026
|
$0.075 | $0.5 | $0.0375 | 262K | Sep 2026 | ||
|
Z
GLM 5.3 Flash is Z.ai's native multimodal model with a 1M-token context window, built on hybrid sparse and linear attention architecture for efficient inference.
$0.15/$0.5I/O
1.05M ctx
Aug 2026
|
$0.15 | $0.5 | $0.03 | 1.05M | Aug 2026 | ||
|
T
Rocinante 12B is a 12-billion-parameter model from TheDrummer with a 65K token context window.
$0.25/$0.5I/O
65K ctx
Sep 2024
|
$0.25 | $0.5 | — | 65K | Sep 2024 | ||
| $0.12 | $0.5 | — | 40K | Apr 2025 | |||
|
T
Cydonia 24B V4.1 is a 24-billion-parameter model from TheDrummer with a 131K token context window.
$0.3/$0.5I/O
131K ctx
Sep 2025
|
$0.3 | $0.5 | $0.15 | 131K | Sep 2025 | ||
|
A
Olmo 3 32B Think is a 32-billion-parameter language model from Allen Institute for AI with an extended reasoning capability and 65k token context window.
$0.15/$0.5I/O
65K ctx
Nov 2025
|
$0.15 | $0.5 | — | 65K | Nov 2025 | ||
| $0.15 | $0.47 | $0.016 | 1M | Sep 2026 | |||
| $0.15 | $0.47 | $0.016 | 1M | Aug 2026 | |||
| $0.117 | $0.455 | — | 131K | Apr 2025 | |||
| $0.15 | $0.45 | $0.015 | 256K | Aug 2025 | |||
| $0.14 | $0.42 | $0.0042 | 1.05M | Sep 2026 | |||
| $0.28 | $0.42 | $0.028 | 131K | Dec 2025 | |||
| $0.28 | $0.42 | $0.028 | 128K | Dec 2024 | |||
| $0.28 | $0.42 | $0.028 | 128K | Jan 2025 | |||
| $0.104 | $0.416 | — | 131K | Oct 2025 | |||
| $0.27 | $0.41 | — | 163K | Sep 2025 | |||
|
P
Laguna M.1 is a mid-sized coding model from Poolside with a 262k token context window.
$0.2/$0.4I/O
262K ctx
Apr 2026
|
$0.2 | $0.4 | $0.1 | 262K | Apr 2026 | ||
| $0.4 | $0.4 | — | 131K | Jul 2024 | |||
| $0.36 | $0.4 | — | 32K | Sep 2024 | |||
|
T
UnslopNemo 12B is a 12-billion-parameter model from TheDrummer with a 1-million-token context window.
$0.4/$0.4I/O
1.02M ctx
Nov 2024
|
$0.4 | $0.4 | — | 1.02M | Nov 2024 | ||
| $0.1 | $0.4 | $0.01 | 1.05M | Jul 2025 | |||
| $0.05 | $0.4 | $0.005 | 400K | Aug 2025 | |||
|
N
Hermes 4 70B is a 70-billion parameter instruction-tuned model from Nous with a 131K token context window.
$0.13/$0.4I/O
131K ctx
Aug 2025
|
$0.13 | $0.4 | — | 131K | Aug 2025 | ||
|
N
A 49-billion-parameter model from NVIDIA with a 131K token context window, positioned between the earlier Nemotron 3 Super and the newer mixture-of-experts models in NVIDIA's lineup.
$0.4/$0.4I/O
131K ctx
Oct 2025
|
$0.4 | $0.4 | — | 131K | Oct 2025 | ||
|
Z
GLM 4.7 Flash is Z.ai's lightweight inference variant of the GLM 4.7 foundation model, designed for faster execution with a 131K-token context window.
$0.0605/$0.4I/O
131K ctx
Jan 2026
|
$0.0605 | $0.4 | — | 131K | Jan 2026 | ||
|
B
A lightweight general-purpose model from ByteDance Seed with a 262k token context window.
$0.1/$0.4I/O
262K ctx
Feb 2026
|
$0.1 | $0.4 | — | 262K | Feb 2026 | ||
| $0.2 | $0.4 | — | 262K | Apr 2026 | |||
| $0.1 | $0.4 | $0.025 | 1M | Apr 2025 | |||
|
I
Mercury 3 is a diffusion-based language model from Inception, and the lab's most capable model to date. It is suited for agentic workflows and real-time applications where latency compounds, such...
$0.1/$0.35I/O
1M ctx
Oct 2026
|
$0.1 | $0.35 | $0.01 | 1M | Oct 2026 | ||
| $0.08 | $0.35 | $0.04 | 131K | Mar 2025 | |||
|
M
A smaller, instruction-tuned variant of Phi 4 designed to run efficiently on resource-constrained hardware while handling a 128K token context.
$0.08/$0.35I/O
128K ctx
Oct 2025
|
$0.08 | $0.35 | $0.08 | 128K | Oct 2025 | ||
| $0.345 | $0.345 | — | 131K | Sep 2024 | |||
| $0.11 | $0.33 | $0.0035 | 1.05M | Aug 2026 | |||
| $0.11 | $0.33 | $0.0035 | 1.05M | Jul 2026 | |||
| $0.05 | $0.33 | — | 131K | Sep 2024 | |||
| $0.1 | $0.32 | — | 131K | Dec 2024 | |||
| $0.075 | $0.3 | $0.0075 | 262K | Mar 2026 | |||
| $0.075 | $0.3 | $0.0375 | 128K | Jul 2024 | |||
| $0.1 | $0.3 | — | 262K | Jul 2025 | |||
| $0.075 | $0.3 | $0.0375 | 131K | Oct 2025 | |||
| $0.1 | $0.3 | $0.01 | 32K | Oct 2025 | |||
|
X
MiMo-V2-Flash is Xiaomi's lightweight language model with a 262K token context window, positioned as an earlier generation model before the MiMo-V2.5 and V2.6 lineups.
$0.1/$0.3I/O
262K ctx
Dec 2025
|
$0.1 | $0.3 | $0.01 | 262K | Dec 2025 | ||
|
B
Seed 1.6 Flash is a lightweight variant of the Seed 1.6 general-purpose model from ByteDance Seed, optimized for faster inference with a 262k token context window.
$0.075/$0.3I/O
262K ctx
Dec 2025
|
$0.075 | $0.3 | — | 262K | Dec 2025 | ||
|
S
Step 3.5 Flash is a lightweight language model from StepFun with a 262K token context window, designed for fast inference on tasks that don't require the full capacity of larger models.
$0.1/$0.3I/O
262K ctx
Jan 2026
|
$0.1 | $0.3 | — | 262K | Jan 2026 | ||
| $0.09 | $0.3 | $0.05 | 262K | Apr 2026 | |||
| $0.1 | $0.3 | — | 262K | Mar 2026 | |||
| $0.08 | $0.3 | — | 1M | Apr 2025 | |||
|
T
Hy-MT2-7B is a 7-billion-parameter translation model from Tencent supporting multilingual translation across dozens of language pairs.
$0.074/$0.295I/O
8K ctx
Aug 2026
|
$0.074 | $0.295 | — | 8K | Aug 2026 | ||
|
T
Hy-MT2-30B-A3B is a 30-billion-parameter translation model from Tencent.
$0.074/$0.295I/O
8K ctx
Aug 2026
|
$0.074 | $0.295 | — | 8K | Aug 2026 | ||
| $0.29 | $0.29 | — | 32K | Jan 2025 | |||
|
X
MiMo-V2.6-Flash is a text language model from Xiaomi built on a Mixture-of-Experts architecture with 309B total parameters and 15B activated per token, supporting a 1M token context window.
$0.14/$0.28I/O
1.05M ctx
Sep 2026
|
$0.14 | $0.28 | $0.0028 | 1.05M | Sep 2026 | ||
| $0.08 | $0.28 | — | 40K | Apr 2025 | |||
| $0.07 | $0.28 | — | 262K | Jul 2025 | |||
| $0.14 | $0.28 | $0.0028 | 1M | Apr 2026 | |||
| $0.065 | $0.26 | — | 1M | Feb 2026 | |||
| $0.05 | $0.25 | $0.005 | 1M | Oct 2026 | |||
| $0.05 | $0.25 | $0.005 | 1.05M | Sep 2026 | |||
| $0.05 | $0.25 | $0.005 | 1.05M | Sep 2026 | |||
|
N
An agentic model from Nex AGI designed to convert goals into executable, verified outcomes with a 262K token context window.
$0.075/$0.25I/O
262K ctx
Sep 2026
|
$0.075 | $0.25 | $0.015 | 262K | Sep 2026 | ||
|
I
Granite 4.2 8B is an 8-billion-parameter dense reasoning model from IBM with a 131K-token context window, designed for tasks requiring multi-step reasoning.
$0.06/$0.25I/O
131K ctx
Aug 2026
|
$0.06 | $0.25 | $0.015 | 131K | Aug 2026 | ||
| $0.17 | $0.25 | — | 262K | Mar 2026 | |||
| $0.17 | $0.25 | — | 262K | Mar 2026 | |||
|
A
A text-only language model from Amazon's Nova family with a 300K-token context window, released in December 2024.
$0.06/$0.24I/O
300K ctx
Dec 2024
|
$0.06 | $0.24 | — | 300K | Dec 2024 | ||
| $0.12 | $0.24 | — | 40K | Apr 2025 | |||
|
N
A 30-billion-parameter mixture-of-experts model from NVIDIA with 3 billion active parameters per token and 262K token context window.
$0.06/$0.24I/O
262K ctx
Dec 2025
|
$0.06 | $0.24 | — | 262K | Dec 2025 | ||
|
I
A 3B-parameter model for extracting structured data from HTML and unstructured text into JSON format using JSON schema instructions.
$0.05/$0.23I/O
128K ctx
Sep 2026
|
$0.05 | $0.23 | $0.05 | 128K | Sep 2026 | ||
|
I
Ling 3.1 Flash is a hybrid reasoning mixture-of-experts model from inclusionAI, with 25B active parameters out of 560B total.
$0.075/$0.225I/O
262K ctx
Oct 2026
|
$0.075 | $0.225 | $0.015 | 262K | Oct 2026 | ||
| $0.027 | $0.201 | — | 60K | Sep 2024 | |||
|
U
Solar Mini 4 is a 35-parameter mixture-of-experts model from Upstage with 3B active parameters and a 524K token context window.
$0.05/$0.2I/O
524K ctx
Sep 2026
|
$0.05 | $0.2 | $0.005 | 524K | Sep 2026 | ||
|
N
A 4-billion-parameter multimodal guardrail model from NVIDIA designed to moderate both inputs to and outputs from language and vision models.
$0.2/$0.2I/O
131K ctx
Jun 2026
|
$0.2 | $0.2 | — | 131K | Jun 2026 | ||
| $0.1 | $0.2 | $0.002 | 1.05M | Sep 2026 | |||
| $0.1 | $0.2 | $0.002 | 1.05M | Aug 2026 | |||
|
N
A 30-billion-parameter mixture-of-experts model from NVIDIA with 3B active parameters per token and a 262K token context window.
$0.07/$0.2I/O
262K ctx
Aug 2026
|
$0.07 | $0.2 | $0.04 | 262K | Aug 2026 | ||
| $0.05 | $0.2 | $0.0125 | 1.05M | Apr 2025 | |||
| $0.05 | $0.2 | $0.01 | 1.05M | Jul 2025 | |||
| $0.025 | $0.2 | $0.0025 | 400K | Aug 2025 | |||
|
P
Laguna XS.2 is a lightweight coding model from Poolside with a 262k token context window.
$0.1/$0.2I/O
262K ctx
Apr 2026
|
$0.1 | $0.2 | $0.05 | 262K | Apr 2026 | ||
| $0.1 | $0.2 | — | 32K | Oct 2024 | |||
|
R
Reka Flash 3 is a multimodal model from Rekaai with a 65K token context window, positioned as a mid-tier offering between the smaller Edge variant and larger models in the Reka lineup.
$0.1/$0.2I/O
65K ctx
Mar 2025
|
$0.1 | $0.2 | — | 65K | Mar 2025 | ||
| $0.075 | $0.2 | — | 128K | Jun 2025 | |||
|
B
UI-TARS 7B is a 7-billion-parameter language model from ByteDance designed for understanding and reasoning about user interface interactions.
$0.1/$0.2I/O
128K ctx
Jul 2025
|
$0.1 | $0.2 | $0.1 | 128K | Jul 2025 | ||
| $0.2 | $0.2 | $0.02 | 262K | Dec 2025 | |||
|
I
Ling 3.0 Flash VL is a multimodal instruct model from InclusionAI that combines efficient language processing with native visual perception capabilities.
$0.06/$0.18I/O
131K ctx
Sep 2026
|
$0.06 | $0.18 | $0.012 | 131K | Sep 2026 | ||
|
Z
GLM 5.3 (batch) is Z.ai's reasoning model optimized for asynchronous batch processing with a 1M-token context window.
$0.05/$0.18I/O
1.05M ctx
Aug 2026
|
$0.05 | $0.18 | $0.01 | 1.05M | Aug 2026 | ||
|
P
Laguna S 2.1 is a 118B parameter coding agent model from Poolside with a 1M token context window, designed for extended code understanding and generation tasks.
$0.09/$0.18I/O
1.05M ctx
Jul 2026
|
$0.09 | $0.18 | $0.009 | 1.05M | Jul 2026 | ||
| $0.18 | $0.18 | — | 163K | Apr 2025 | |||
|
T
Hy-MT2-1.8B is a 1.8-billion-parameter translation model from Tencent supporting 33 language pairs and Chinese dialects and minority languages.
$0.044/$0.177I/O
8K ctx
Aug 2026
|
$0.044 | $0.177 | — | 8K | Aug 2026 | ||
| $0.05 | $0.15 | $0.004 | 1.05M | Sep 2026 | |||
|
I
A 3B-parameter model from Inference.net designed to extract structured data from HTML and unstructured text into JSON format using schema-based instructions.
$0.03/$0.15I/O
128K ctx
Sep 2026
|
$0.03 | $0.15 | $0.03 | 128K | Sep 2026 | ||
|
I
Mercury 2.5 is a diffusion-based language model from Inception that generates and refines multiple tokens in parallel instead of sequentially.
$0.04/$0.15I/O
260K ctx
Sep 2026
|
$0.04 | $0.15 | $0.004 | 260K | Sep 2026 | ||
| $0.0375 | $0.15 | — | 128K | Dec 2024 | |||
| $0.05 | $0.15 | — | 131K | Mar 2025 | |||
|
A
Trinity Mini is a smaller reasoning model from Arcee AI with a 131k token context window for handling intermediate problem-solving and reasoning tasks.
$0.045/$0.15I/O
131K ctx
Dec 2025
|
$0.045 | $0.15 | — | 131K | Dec 2025 | ||
| $0.15 | $0.15 | $0.015 | 262K | Dec 2025 | |||
|
E
An instruction-tuned language model from EssentialAI with a 32K token context window.
$0.15/$0.15I/O
32K ctx
Dec 2025
|
$0.15 | $0.15 | — | 32K | Dec 2025 | ||
| $0.14 | $0.14 | — | 8K | Apr 2024 | |||
|
A
Nova Micro 1.0 is Amazon's smallest and fastest text-only language model from the Nova family, designed for low-latency applications with minimal computational overhead.
$0.035/$0.14I/O
128K ctx
Dec 2024
|
$0.035 | $0.14 | — | 128K | Dec 2024 | ||
|
M
A 14-billion-parameter language model from Microsoft designed for general-purpose text generation and reasoning tasks with a 16K token context window.
$0.07/$0.14I/O
16K ctx
Jan 2025
|
$0.07 | $0.14 | — | 16K | Jan 2025 | ||
| $0.0296 | $0.136 | — | 131K | Aug 2025 | |||
| $0.03 | $0.13 | $0.006 | 1M | Jul 2026 | |||
|
I
Ling 3.0 Flash Sante is a health and medicine-focused mixture-of-experts model from InclusionAI, built on Ling 3.0 Flash with 5.1B active parameters out of 124B total. It is designed for...
$0.042/$0.1232I/O
262K ctx
Sep 2026
|
$0.042 | $0.1232 | $0.0084 | 262K | Sep 2026 | ||
|
I
Ling 3.0 Flash Fin is a fast, efficient instruct model from inclusionAI with a 262K-token context window.
$0.042/$0.1232I/O
262K ctx
Aug 2026
|
$0.042 | $0.1232 | $0.0084 | 262K | Aug 2026 | ||
|
P
Laguna XS 2.1 is a lightweight coding model from Poolside with a 262k token context window.
$0.06/$0.12I/O
262K ctx
Jul 2026
|
$0.06 | $0.12 | $0.03 | 262K | Jul 2026 | ||
| $0.06 | $0.12 | — | 32K | May 2025 | |||
|
L
A 24-billion-parameter open-weight model from LiquidAI with a 32K token context window.
$0.03/$0.12I/O
32K ctx
Feb 2026
|
$0.03 | $0.12 | — | 32K | Feb 2026 | ||
| $0.024 | $0.112 | — | 131K | Aug 2025 | |||
|
I
A lightweight language model from IBM's Granite family with a 131K-token context window, designed for efficient inference on resource-constrained environments.
$0.017/$0.112I/O
131K ctx
Oct 2025
|
$0.017 | $0.112 | — | 131K | Oct 2025 | ||
| $0.11 | $0.11 | — | 128K | Oct 2024 | |||
|
G
MythoMax 13B is a 13-billion-parameter model from Gryphe designed for creative and narrative tasks with a 4096-token context window.
$0.08/$0.11I/O
4K ctx
Jul 2023
|
$0.08 | $0.11 | — | 4K | Jul 2023 | ||
|
N
An agentic model from Nex AGI designed to convert goals into executable, verified outcomes with a 262K token context window.
$0.025/$0.1I/O
262K ctx
Sep 2026
|
$0.025 | $0.1 | $0.0025 | 262K | Sep 2026 | ||
|
N
An earlier iteration from Nex AGI with a 262K token context window, positioned between the N2-Pro and the newer N2.5 agentic models.
$0.025/$0.1I/O
262K ctx
Jun 2026
|
$0.025 | $0.1 | $0.0025 | 262K | Jun 2026 | ||
| $0.05 | $0.1 | — | 131K | Mar 2025 | |||
| $0.1 | $0.1 | $0.01 | 131K | Dec 2025 | |||
|
R
Reka Edge is a smaller instruction-tuned model from Reka with a 16K token context window.
$0.1/$0.1I/O
16K ctx
Mar 2026
|
$0.1 | $0.1 | — | 16K | Mar 2026 | ||
|
I
An 8-billion-parameter language model from IBM's Granite family with a 131K-token context window.
$0.05/$0.1I/O
131K ctx
Apr 2026
|
$0.05 | $0.1 | $0.05 | 131K | Apr 2026 | ||
| $0.018 | $0.09 | $0.009 | 131K | Aug 2025 | |||
| $0.05 | $0.08 | $0.025 | 131K | Jul 2024 | |||
| $0.05 | $0.08 | — | 32K | Jan 2025 | |||
| $0.075 | $0.075 | $0.0075 | 262K | Dec 2025 | |||
|
I
Ling-3.0-flash is a fast, efficient instruct model from inclusionAI with a 262K-token context window.
$0.021/$0.063I/O
262K ctx
Jul 2026
|
$0.021 | $0.063 | $0.0042 | 262K | Jul 2026 | ||
|
S
A smaller, earlier-generation 8-billion-parameter model from Sao10K based on Llama 3, with an 8192-token context window.
$0.04/$0.05I/O
8K ctx
Aug 2024
|
$0.04 | $0.05 | — | 8K | Aug 2024 | ||
| $0.029 | $0.03 | — | 131K | Jul 2024 | |||
| $0.484 | $0.03 | — | 131K | Feb 2025 | |||
|
I
Ling-2.6-flash is a fast, efficient instruct model from inclusionAI with a 262K-token context window.
$0.01/$0.03I/O
262K ctx
Apr 2026
|
$0.01 | $0.03 | $0.002 | 262K | Apr 2026 | ||
|
S
Extract is a description-driven entity extraction model from ScaleDown. Instead of a fixed set of entity types, callers describe the fields they want in plain English through `provider.options.scaledown.entities`, and the...
$0.04/$0I/O
1M ctx
Oct 2026
|
$0.04 | $0 | — | 1M | Oct 2026 | ||
|
S
Summarize is an abstractive summarization model from ScaleDown. It condenses long text and documents into a fluent, faithful rewrite rather than lifting sentences from the source, and is suited for...
$0.04/$0I/O
1M ctx
Oct 2026
|
$0.04 | $0 | — | 1M | Oct 2026 |
Sorted by output price by default — output tokens are typically 3–5× more expensive than input and dominate the bill for generation-heavy workloads. Sort by input or cached input when your mix differs.