New: Agent Index for knowledge work is live.Explore the index ↗
Compare models

LLM leaderboard for open sources models - 2026

Compare models using publicly reported benchmark results and find the right one for reasoning, coding, tool use, browsing, personal assistance, and everyday tasks.

Updated: Data as of:

Best Overall

Based on Humanity’s Last Exam, one of the hardest AI benchmarks.

Humanity’s Last Exam rankings as JSON
  1. 70.0K
    Kimi K3.1 ProRank 1: Kimi K3.1 Pro by Moonshot AI, 70 percent. With tools. Stable HLE — full set. Max effort; temperature=1.0; top_p=0.95; full 2500-question set.. Reported result source for Kimi K3.1 Pro
  2. 63.9D
    DeepSeek V4.1 FlashRank 2: DeepSeek V4.1 Flash by DeepSeek, 63.9 percent. With tools. HLE — source does not specify subset. . Reported result source for DeepSeek V4.1 Flash
  3. 62.5Z
    GLM-5.3Rank 3: GLM-5.3 by Z.ai, 62.5 percent. With tools. HLE — source does not specify subset. temperature=1.0; top_p=0.95; max generation 163840 tokens; max context 300000 tokens; context management; GPT-5.6 Luna medium judge.. Reported result source for GLM-5.3
  4. 60.0D
    DeepSeek V4 Pro 0813Rank 4: DeepSeek V4 Pro 0813 by DeepSeek, 60 percent. With tools. HLE — source does not specify subset. . Reported result source for DeepSeek V4 Pro 0813
  5. 59.8K
    Kimi K3Rank 5: Kimi K3 by Moonshot AI, 59.8 percent. With tools. HLE — source does not specify subset. temperature=1.0; top_p=0.95; max generation 163840 tokens; max context 300000 tokens; context management; GPT-5.6 Luna medium judge.. Reported result source for Kimi K3
  6. 58.8D
    DeepSeek V4.1 Flash 1017Rank 6: DeepSeek V4.1 Flash 1017 by DeepSeek, 58.8 percent. With tools. Stable HLE — full set. Max effort; competitor result in Moonshot AI report.. Reported result source for DeepSeek V4.1 Flash 1017
  7. 55.3Z
    GLM-5.3-FlashRank 7: GLM-5.3-Flash by Z.ai, 55.3 percent. With tools. Stable HLE — full set. Primary bench_53.png; full set; temperature 1; top_p .95; 163840 generation; 300K context; GPT-5.6 Luna judge. Reported result source for GLM-5.3-Flash
  8. 54.7Z
    GLM-5.2Rank 8: GLM-5.2 by Z.ai, 54.7 percent. With tools. Stable HLE — text-only subset. Text-only per model-card footnote; 163840 generation; 300K context; no context management; GPT-5.5 medium judge. Reported result source for GLM-5.2
  9. 54.0K
    Kimi K2.6Rank 9: Kimi K2.6 by Moonshot AI, 54 percent. With tools. HLE — source does not specify subset. Provider announcement; tools enabled; subset not explicitly named in announcement. Reported result source for Kimi K2.6
  10. 51.0K
    Kimi K2 ThinkingRank 10: Kimi K2 Thinking by Moonshot AI, 51 percent. With tools. Stable HLE — text-only subset. Heavy configuration: parallel test-time compute. Text-only; retained separately from single-agent 44.9.. Reported result source for Kimi K2 Thinking
  11. 50.2K
    Kimi K2.5Rank 11: Kimi K2.5 by Moonshot AI, 50.2 percent. With tools. Stable HLE — full set. Full text and image set; search, code and browsing; 96K completion; simple context management; Hugging Face blocked. Reported result source for Kimi K2.5
  12. 49.4?
    MiMo-V2.6-ProRank 12: MiMo-V2.6-Pro by Xiaomi, 49.4 percent. Without tools. Stable HLE — text-only subset. MiMo-V2.6-Pro; original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal.. Reported result source for MiMo-V2.6-Pro
  13. 48.2D
    DeepSeek V4 Pro 0424Rank 13: DeepSeek V4 Pro 0424 by DeepSeek, 48.2 percent. With tools. HLE — source does not specify subset. Original V4 Think Max; HLE with tools; source does not explicitly identify subset. Reported result source for DeepSeek V4 Pro 0424
  14. 48.0?
    MiniMax M3.1 FlashRank 14: MiniMax M3.1 Flash by MiniMax, 48 percent. Without tools. Stable HLE — text-only subset. MiniMax-M3.1-Flash; original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal.. Reported result source for MiniMax M3.1 Flash
  15. 46.5?
    Step 5 PreviewRank 15: Step 5 Preview by StepFun, 46.5 percent. Without tools. Stable HLE — text-only subset. Step 5 Preview; original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal.. Reported result source for Step 5 Preview
  16. 45.1D
    DeepSeek V4 Flash 0420Rank 16: DeepSeek V4 Flash 0420 by DeepSeek, 45.1 percent. With tools. HLE — source does not specify subset. Original V4 Think Max; HLE with tools; source does not explicitly identify subset. Reported result source for DeepSeek V4 Flash 0420
  17. 43.9M
    Muse Spark OpenRank 17: Muse Spark Open by Meta, 43.9 percent. Without tools. Stable HLE — text-only subset. Muse Spark Open (xhigh); original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal.. Reported result source for Muse Spark Open
  18. 40.8D
    DeepSeek V3.2Rank 18: DeepSeek V3.2 by DeepSeek, 40.8 percent. With tools. Stable HLE — text-only subset. Published comparison table; DeepSeek dagger denotes text-only. Other rows use full text and image set.. Reported result source for DeepSeek V3.2
  19. 39.0?
    MiniMax M3Rank 19: MiniMax M3 by MiniMax, 38.9712696941613 percent. Without tools. Stable HLE — text-only subset. MiniMax-M3; original HLE text-only 2,158 questions; no tools; not HLE Diamond. Reported result source for MiniMax M3
  20. 33.9?
    Qwen3.8 27BRank 20: Qwen3.8 27B by Qwen, 33.9 percent. Without tools. Stable HLE — text-only subset. Qwen3.8 27B (xhigh); original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal.. Reported result source for Qwen3.8 27B

Best models per task

Benchmark rankings as JSON

Best in Reasoning

GPQA Diamond
  1. Kimi K3 by Moonshot AI93.5%
    Kimi K3 (Max); 198-question Diamond subset; independent AA evaluation. Source for Kimi K3 on GPQA Diamond
  2. MiniMax M3 by MiniMax92.9%
    MiniMax-M3; 198-question Diamond subset; independent AA evaluation. Source for MiniMax M3 on GPQA Diamond
  3. DeepSeek V4 Pro 0813 by DeepSeek92.8%
    DeepSeek V4 Pro 0813 (Max); 198-question Diamond subset; independent AA evaluation. Source for DeepSeek V4 Pro 0813 on GPQA Diamond
  4. GLM-5.3 by Z.ai91.7%
    GLM-5.3 (Max); 198-question Diamond subset; independent AA evaluation. Source for GLM-5.3 on GPQA Diamond
  5. GLM-5.3-Flash by Z.ai91.2%
    GLM-5.3-Flash; 198-question Diamond subset; independent AA evaluation. Source for GLM-5.3-Flash on GPQA Diamond

Best in Coding

SWE-bench Verified
  1. DeepSeek V4 Pro 0424 by DeepSeek80.6%
    DeepSeek original V4 release; Think Max; published SWE Verified resolved score; original release, not 0813. Source for DeepSeek V4 Pro 0424 on SWE-bench Verified
  2. MiniMax M3 by MiniMax80.5%
    MiniMax primary benchmark table; internal Claude Code harness; system prompt override; average of four runs. Source for MiniMax M3 on SWE-bench Verified
  3. Kimi K2.6 by Moonshot AI80.2%
    Kimi K2.6 Thinking; quoted comparison table; follow original Kimi SWE harness. Source for Kimi K2.6 on SWE-bench Verified
  4. DeepSeek V4 Flash 0420 by DeepSeek79.0%
    DeepSeek original V4 release; Think Max; published SWE Verified resolved score; original release, not 0813. Source for DeepSeek V4 Flash 0420 on SWE-bench Verified
  5. Kimi K2.5 by Moonshot AI76.8%
    Internal minimal coding agent; non-thinking; average of five independent runs. Source for Kimi K2.5 on SWE-bench Verified

Best in Terminal Use

Terminal-Bench 2.1
  1. Kimi K3 by Moonshot AI85.0%
    Kimi K3 (Max); 89 tasks; Terminus 2; e2b; three repeats. Source for Kimi K3 on Terminal-Bench 2.1
  2. GLM-5.3-Flash by Z.ai84.3%
    GLM-5.3-Flash; 89 tasks; Terminus 2; e2b; three repeats. Source for GLM-5.3-Flash on Terminal-Bench 2.1
  3. GLM-5.3 by Z.ai83.9%
    GLM-5.3 (Max); 89 tasks; Terminus 2; e2b; three repeats. Source for GLM-5.3 on Terminal-Bench 2.1
  4. DeepSeek V4 Pro 0813 by DeepSeek78.7%
    DeepSeek V4 Pro 0813 (Max); 89 tasks; Terminus 2; e2b; three repeats. Source for DeepSeek V4 Pro 0813 on Terminal-Bench 2.1
  5. GLM-5.2 by Z.ai77.9%
    GLM-5.2 (Max); 89 tasks; Terminus 2; e2b; three repeats. Source for GLM-5.2 on Terminal-Bench 2.1

Best in Web Research

BrowseComp
  1. Kimi K3.1 Pro by Moonshot AI92.1%
    200K context compaction; max 500 steps; temperature 1; top-p 1; preserved original track kimi31-report-200k-500steps. Source for Kimi K3.1 Pro on BrowseComp
  2. Kimi K3 by Moonshot AI91.2%
    Max effort; temperature 1; top_p 1; context compaction triggered at 300K tokens. Primary General Agents benchmark image. Separate no-management 1M track scores 90.4.. Source for Kimi K3 on BrowseComp
  3. MiniMax M3 by MiniMax83.5%
    Primary full table; WebExplorer-based agent; discard history after 64K tokens. Source for MiniMax M3 on BrowseComp
  4. DeepSeek V4 Pro 0424 by DeepSeek83.4%
    Think Max; provider browsing agent; discard-all context management; original V4 release. Source for DeepSeek V4 Pro 0424 on BrowseComp
  5. Kimi K2.6 by Moonshot AI83.2%
    Moonshot provider-reported tool-augmented BrowseComp. Source for Kimi K2.6 on BrowseComp

Best in Work Automations

AutomationBench-AA
  1. Kimi K3.1 Pro by Moonshot AI73.30%
    Artificial Analysis evaluation setup. Source for Kimi K3.1 Pro on AutomationBench-AA
  2. MiniMax M3.1 Flash by MiniMax72.52%
    MiniMax-M3.1-Flash; Private 657 tasks; 50 turns; objective completion. Source for MiniMax M3.1 Flash on AutomationBench-AA
  3. DeepSeek V4.1 Flash by DeepSeek68.89%
    DeepSeek V4.1 Flash (Max); Private 657 tasks; 50 turns; objective completion. Source for DeepSeek V4.1 Flash on AutomationBench-AA
  4. DeepSeek V4.1 Flash 1017 by DeepSeek68.49%
    DeepSeek V4.1 Flash 1017 (High); Private 657 tasks; 50 turns; objective completion. Source for DeepSeek V4.1 Flash 1017 on AutomationBench-AA
  5. GLM-5.3 by Z.ai62.20%
    GLM-5.3 (Max); Private 657 tasks; 50 turns; objective completion. Source for GLM-5.3 on AutomationBench-AA

Best in Knowledge Work

GDPval-AA v2.1
  1. Kimi K3.1 Pro by Moonshot AI1,858 Elo
    Kimi K3.1 Pro; 220 tasks; Stirrup agent; pairwise Elo. Source for Kimi K3.1 Pro on GDPval-AA v2.1
  2. DeepSeek V4.1 Flash 1017 by DeepSeek1,746 Elo
    DeepSeek V4.1 Flash 1017 (Max); 220 tasks; Stirrup agent; pairwise Elo. Source for DeepSeek V4.1 Flash 1017 on GDPval-AA v2.1
  3. GLM-5.3 by Z.ai1,657 Elo
    GLM-5.3 (Max); 220 tasks; Stirrup agent; pairwise Elo. Source for GLM-5.3 on GDPval-AA v2.1
  4. GLM-5.3-Flash by Z.ai1,649 Elo
    GLM-5.3-Flash; 220 tasks; Stirrup agent; pairwise Elo. Source for GLM-5.3-Flash on GDPval-AA v2.1
  5. DeepSeek V4.1 Flash by DeepSeek1,600 Elo
    DeepSeek V4.1 Flash (Max); 220 tasks; Stirrup agent; pairwise Elo. Source for DeepSeek V4.1 Flash on GDPval-AA v2.1

Fastest and most affordable models

Best reported provider per model for each metric.

Speed, latency and price rankings as JSON

Fastest models

tokens/sec

Output speed · higher is better

  1. DeepSeek V4.1 Flash 1017 by DeepSeekvia Inco (FAST)1,075.3DeepSeek V4.1 Flash 1017 (max) (FAST) · Artificial Analysis: DeepSeek V4.1 Flash 1017 (max) (FAST); host Inco (FAST); standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  2. DeepSeek V4.1 Flash by DeepSeekvia Baseten (FAST)830DeepSeek V4.1 Flash (max) (FAST) · Artificial Analysis: DeepSeek V4.1 Flash (max) (FAST); host Baseten (FAST); standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  3. MiniMax M3.1 Flash by MiniMaxvia Inco812.6MiniMax-M3.1-Flash · Artificial Analysis: MiniMax-M3.1-Flash; host Inco; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  4. GLM-5.3 by Z.aivia Systalyze688.2GLM-5.3 (max) · Artificial Analysis: GLM-5.3 (max); host Systalyze; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  5. Kimi K3 by Moonshot AIvia Inco (FAST)481.7Kimi K3 (max) (FAST) · Artificial Analysis: Kimi K3 (max) (FAST); host Inco (FAST); standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.

Lowest latency

TTFT

Time to first token · lower is better

  1. DeepSeek V4.1 Flash 1017 by DeepSeekvia Baseten (FAST)0.182sDeepSeek V4.1 Flash 1017 (max) (FAST) · Artificial Analysis: DeepSeek V4.1 Flash 1017 (max) (FAST); host Baseten (FAST); standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  2. DeepSeek V4.1 Flash by DeepSeekvia Together AI0.311sDeepSeek V4.1 Flash (max) · Artificial Analysis: DeepSeek V4.1 Flash (max); host Together AI; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  3. MiniMax M3.1 Flash by MiniMaxvia Together AI0.584sMiniMax-M3.1-Flash · Artificial Analysis: MiniMax-M3.1-Flash; host Together AI; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  4. GLM-5.3 by Z.aivia Together AI0.597sGLM-5.3 (max) · Artificial Analysis: GLM-5.3 (max); host Together AI; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  5. MiniMax M3 by MiniMaxvia MiniMax0.699sMiniMax-M3 · Explicit first-party AA measurement; reasoning configuration retained. Cross-provider medians are excluded. Retrieval time is not measurement time.

Most affordable

$/1M tokens

50% input + 50% output · lower is better

  1. DeepSeek V4.1 Flash 1017 by DeepSeekvia DeepInfra$0.196Input $0.098/1M · Output $0.294/1M · Artificial Analysis: DeepSeek V4.1 Flash 1017 (max); host DeepInfra; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  2. DeepSeek V4.1 Flash by DeepSeekvia DeepInfra$0.28Input $0.14/1M · Output $0.42/1M · Artificial Analysis: DeepSeek V4.1 Flash (max); host DeepInfra; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  3. GLM-5.3-Flash by Z.aivia Z.ai$0.325Input $0.15/1M · Output $0.5/1M · Explicit first-party AA measurement; reasoning configuration retained. Cross-provider medians are excluded. Retrieval time is not measurement time.
  4. MiniMax M3.1 Flash by MiniMaxvia GMI$0.356Input $0.1425/1M · Output $0.57/1M · Artificial Analysis: MiniMax-M3.1-Flash; host GMI; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.
  5. GLM-5.3 by Z.aivia Multiverse Computing$0.42Input $0.2/1M · Output $0.64/1M · Artificial Analysis: GLM-5.3 (max); host Multiverse Computing; standard uncached input/output pricing. TTFT is the first streamed token, which may be reasoning. Retrieval time is not measurement time; region/concurrency/quantization not specified in this record.

Model comparison

Compare each model and provider offering. Costs are per 1M tokens.

Download CSV
Model specifications and provider-specific token pricing, time to first token, and generation speed. A dash means no published value is recorded.
DeepSeek V4 Pro 0813
Baseten1,000,000$1.32$3.96393,2161.589143.882
DeepSeek V4 Pro 0813
Databricks1,000,000$1.32$3.96393,216——
DeepSeek V4 Pro 0813
DeepInfra1,000,000$1.30$2.60393,2161.12978.412
DeepSeek V4 Pro 0813
DeepSeek1,048,576$1.32$3.96393,2161.79389.952
DeepSeek V4 Pro 0813
DigitalOcean1,048,576$1.32$3.96393,2160.83257.677
DeepSeek V4 Pro 0813
GMI1,048,576$1.056$3.168393,2163.15271.407
DeepSeek V4 Pro 0813
Nebius1,048,576$1.32$3.96393,2161.602172.165
DeepSeek V4 Pro 0813
Novita1,048,576$1.32$3.96393,2162.12372.307
DeepSeek V4 Pro 0813
SiliconFlow1,048,576$1.32$3.96393,2162.08864.823
DeepSeek V4 Pro 0813
Systalyze1,048,576$0.50$2.00393,2161.894305.002
DeepSeek V4.1 Flash
Baseten1,000,000$0.30$1.20—0.63244.981
DeepSeek V4.1 Flash
Baseten (FAST)1,048,576$0.60$2.40—0.94830.034
DeepSeek V4.1 Flash
Bitdeer AI1,000,000$0.144$0.576—1.515289.729
DeepSeek V4.1 Flash
CoreWeave1,000,000$0.20$0.65—0.997297.861
DeepSeek V4.1 Flash
Crusoe1,048,576$0.29$1.20—0.771205.633
DeepSeek V4.1 Flash
Databricks1,000,000$0.30$1.20—0.91384.067
DeepSeek V4.1 Flash
DeepInfra1,000,000$0.14$0.42—0.698164.074
DeepSeek V4.1 Flash
DigitalOcean1,000,000$0.30$1.20—0.82181.897
DeepSeek V4.1 Flash
GMI1,000,000$0.18$0.72—1.813140.702
DeepSeek V4.1 Flash
Makora1,048,576$0.26$1.17—0.662190.722
DeepSeek V4.1 Flash
Modal1,000,000$0.30$1.20—0.559157.789
DeepSeek V4.1 Flash
Nebius1,048,576$0.30$1.20—1.334257.082
DeepSeek V4.1 Flash
Novita524,288$0.30$1.20—1.228136.759
DeepSeek V4.1 Flash
Parasail1,048,576$0.30$1.20—0.63239.17
DeepSeek V4.1 Flash
Scaleway262,144$0.70$2.09—1.41201.347
DeepSeek V4.1 Flash
SiliconFlow1,000,000$0.15$0.60—1.596154.517
DeepSeek V4.1 Flash
Together AI1,048,576$0.30$1.20—0.311373.034
DeepSeek V4.1 Flash
Wafer1,048,576$0.30$1.20—0.563320.504
DeepSeek V4.1 Flash 1017
Baseten1,000,000$0.30$1.20393,2160.355166.806
DeepSeek V4.1 Flash 1017
Baseten (FAST)1,048,576$0.60$2.40393,2160.1821,021.729
DeepSeek V4.1 Flash 1017
Bitdeer AI1,000,000$0.144$0.576393,2161.651294.279
DeepSeek V4.1 Flash 1017
DeepInfra1,000,000$0.098$0.294393,2160.70192.086
DeepSeek V4.1 Flash 1017
DeepSeek1,000,000$0.30$1.20393,2161.277173.207
DeepSeek V4.1 Flash 1017
DigitalOcean1,000,000$0.30$1.20393,2160.503253.956
DeepSeek V4.1 Flash 1017
Fireworks1,000,000$0.30$1.20393,2161.338214.831
DeepSeek V4.1 Flash 1017
GMI1,048,575$0.18$0.72393,2161.448363.705
DeepSeek V4.1 Flash 1017
Inco (FAST)1,048,576$0.60$2.40393,2161.3171,075.275
DeepSeek V4.1 Flash 1017
LithosAI1,048,576$0.15$0.60393,2160.962248.975
DeepSeek V4.1 Flash 1017
LithosAI (FAST)1,048,576$0.25$1.00393,2161.073393.111
DeepSeek V4.1 Flash 1017
LithosAI (ULTRA CHAT)1,048,576$0.35$1.40393,216——
DeepSeek V4.1 Flash 1017
LithosAI (ULTRA)1,048,576$0.35$1.40393,2160.996430.879
DeepSeek V4.1 Flash 1017
Makora1,048,576$0.15$1.20393,2160.94211.519
DeepSeek V4.1 Flash 1017
Modal1,048,576$0.25$1.00393,2160.601233.592
DeepSeek V4.1 Flash 1017
Nebius1,048,576$0.30$1.20393,2161.33391.095
DeepSeek V4.1 Flash 1017
Novita1,048,576$0.30$1.20393,2161.618149.885
DeepSeek V4.1 Flash 1017
Parasail1,000,000$0.30$1.20393,2160.623243.147
DeepSeek V4.1 Flash 1017
PrimaLabs1,048,576$0.24$0.96393,2160.803225.044
DeepSeek V4.1 Flash 1017
SiliconFlow1,048,576$0.15$0.60393,2161.162231.297
DeepSeek V4.1 Flash 1017
Systalyze1,048,576$0.30$1.20393,2160.988845.4
DeepSeek V4.1 Flash 1017
Together AI1,048,576$0.30$1.20393,2160.31382.936
DeepSeek V4.1 Flash 1017
Wafer1,048,576$0.15$0.60393,2160.618205.721
GLM-5.2
Z.ai1,000,000$1.40$4.40—1.77794.073
GLM-5.3
Baseten1,048,576$1.40$4.40131,0722.822106.426
GLM-5.3
Baseten (FAST)1,000,000$2.10$6.60131,0720.795276.567
GLM-5.3
Bitdeer AI1,000,000$0.63$1.98131,0721.588156.031
GLM-5.3
Crusoe (FP4)786,432$1.40$4.40131,0720.723269.762
GLM-5.3
Databricks1,000,000$1.40$4.40131,0720.925154.356
GLM-5.3
DeepInfra1,048,575$0.5625$2.50131,0721.01129.923
GLM-5.3
DigitalOcean1,048,576$1.40$4.40131,0720.83758.931
GLM-5.3
FriendliAI1,000,000$1.26$3.96131,0720.793139.209
GLM-5.3
GMI1,000,000$0.70$2.20131,0722.735125.736
GLM-5.3
Inco163,840$1.40$4.40131,0721.53996.309
GLM-5.3
Inco (FAST)1,000,000$2.80$8.80131,0720.913482.734
GLM-5.3
LithosAI (FP4 ULTRA CHAT)1,048,576$2.10$6.60131,072——
GLM-5.3
LithosAI (FP4)1,048,576$1.05$3.30131,0720.91373.189
GLM-5.3
Makora1,048,576$0.14$4.40131,0721.309121.54
GLM-5.3
Modal1,000,000$1.40$4.40131,0720.682203.893
GLM-5.3
Modular1,048,576$1.40$4.40131,0720.634187.466
GLM-5.3
Multiverse Computing1,000,000$0.20$0.64131,0721.284214.631
GLM-5.3
Nebius (FP4)1,000,000$1.40$4.40131,0721.492238.818
GLM-5.3
Novita1,000,000$1.40$4.40131,0721.93890.783
GLM-5.3
Parasail1,000,000$1.40$4.40131,0720.873217.437
GLM-5.3
Scaleway262,144$2.09$6.38131,0721.268265.356
GLM-5.3
Supernova (FP4, FAST)1,000,000$2.80$8.80131,0721.386561.789
GLM-5.3
Supernova (FP4)1,000,000$0.63$1.98131,0721.814207.47
GLM-5.3
Systalyze1,000,000$0.50$1.00131,0720.947688.217
GLM-5.3
Together AI1,048,575$1.40$4.40131,0720.597188.682
GLM-5.3
Wafer1,000,000$1.40$4.40131,0720.891157.852
GLM-5.3
Zai1,000,000$1.40$4.40131,0721.47788.436
GLM-5.3-Flash
Z.ai1,000,000$0.15$0.50—1.54675.174
Inkling
Thinking Machines—$1.00$4.05———
Kimi K2.6
Moonshot AI256,000$0.95$4.00—2.83440.714
Kimi K3
Amazon1,000,000$3.00$15.001,048,5761.421216.139
Kimi K3
Baseten1,048,576$3.00$15.001,048,5761.30290.056
Kimi K3
Databricks204,800$3.00$15.001,048,5760.87115.162
Kimi K3
DigitalOcean1,048,576$3.00$15.001,048,5761.22430.426
Kimi K3
Fireworks1,048,576$3.00$15.001,048,5760.961113.916
Kimi K3
Inco1,000,000$3.00$15.001,048,5761.778175.272
Kimi K3
Inco (FAST)1,048,576$6.00$30.001,048,5761.659481.694
Kimi K3
Kimi1,048,576$3.00$15.001,048,5765.97764.916
Kimi K3
LithosAI (ULTRA CHAT)1,048,576$5.60$28.001,048,576——
Kimi K3
Makora1,048,576$0.65$15.001,048,5761.13384.476
Kimi K3
Modal1,048,576$3.00$15.001,048,5760.933143.314
Kimi K3
Nebius1,048,576$3.00$15.001,048,5761.047212.032
Kimi K3
Parasail1,048,576$3.00$15.001,048,5761.187165.949
Kimi K3
Together AI1,048,576$3.00$15.001,048,5761.31943.821
Kimi K3.1 Pro
Kimi1,000,000$2.00$10.001,048,5763.42731.249
Kimi K3.1 Pro
Kimi High Speed1,000,000$4.00$20.001,048,5762.68376.93
MiMo-V2.6-Pro
Xiaomi—$0.435$0.87———
MiniMax M3
MiniMax1,000,000$0.30$1.20—0.699119.672
MiniMax M3.1 Flash
DigitalOcean1,048,576$0.15$0.60524,2880.634325.756
MiniMax M3.1 Flash
Fireworks1,000,000$0.15$0.60524,2880.929160.733
MiniMax M3.1 Flash
GMI1,000,000$0.1425$0.57524,2882.268219.719
MiniMax M3.1 Flash
Inco262,144$0.30$1.20524,2881.403812.597
MiniMax M3.1 Flash
Inco1,000,000$0.15$0.60524,2881.84389.294
MiniMax M3.1 Flash
MiniMax1,000,000$0.15$0.60524,2880.904136.2
MiniMax M3.1 Flash
Modular1,000,000$0.15$0.60524,2880.893209.151
MiniMax M3.1 Flash
Nebius1,000,000$0.15$0.60524,2881.723114.21
MiniMax M3.1 Flash
Novita1,000,000$0.15$0.60524,2881.613127.737
MiniMax M3.1 Flash
Parasail1,000,000$0.15$0.60524,2880.849190.503
MiniMax M3.1 Flash
SambaNova1,048,576$0.30$1.20524,2882.213159.122
MiniMax M3.1 Flash
SiliconFlow1,000,000$0.15$0.60524,2882.739151.741
MiniMax M3.1 Flash
Together AI1,048,576$0.15$0.60524,2880.584140.861