LLM leaderboard · 278 models · 57 labs
The LLM leaderboard: who is smartest, who is cheapest, who is fastest.
Gemini 4 Carbon (high) from Google leads the LLM leaderboard at 59.4. Within 5 points of it, GPT-6.1 Sol (max) runs the full Intelligence Index for $1.27, 84% less than the leader. K2 Horizon 375B A23B is the fastest scored model at 720 tokens per second, and the best open-weights model is DeepSeek V4.1 Flash 1017 (max) at 32.5.
Which LLM should you pick? The best model for each job
Six answers computed from the data below, so they update when the data does. Each card links to the lab that makes the model.
- Most intelligentGemini 4 Carbon (high)59.4 index
1 point ahead of Claude Opus 5.5 (max with fallback). Costs $7.76 to run the full index.
Google · Google DeepMind - Best value near the topGPT-6.1 Sol (max)$1.27 per index run
Scores 54.5, 4.9 points below the leader, for 84% less than the leader's $7.76.
OpenAI · OpenAI - Fastest scored modelK2 Horizon 375B A23B720 tokens/s
Scores 17.8, 41.7 points below the leader: speed here costs capability.
Institute of Foundation Models - Best open weightsDeepSeek V4.1 Flash 1017 (max)32.5 index
26.9 points below the leader, and the weights are published. 34 of the 90 scored models are open.
DeepSeek · DeepSeek - Longest context, scoredMuse Wonder (max)1M tokens
Scores 43.3. Several scored models share this window; this is the strongest of them.
Meta · AI at Meta - Cheapest output near the topGPT-6.1 Terra (max)$5.00 per 1M output
Among models within 15 points of the leader, the lowest output price. Scores 51.0.
OpenAI · OpenAI
Our take. For most work, start with the best-value pick and move up to the leader only if your own tests show the gap matters. The top models sit close together, so price and speed decide more purchases than the last few index points. For which companies sit behind these models, see the top 10 AI companies.
Compare all 278 LLMs: filter, sort and chart
Filters apply to every chart and table below. Select any model to see its price, speed and benchmark profile against the leader.
Showing 90 of 278 models · 90 with an Intelligence Index score
Gemini 4 Carbon (high)
Google · released 2026-10-28 · closed weights · reasoning
- Index cost
- $7.76
- Input / 1M
- $4.00
- Output / 1M
- $20.00
- Tokens / s
- 95
- Latency
- 137.6 s
- Context
- 1M
Benchmarks (%). The tick marks the leader, Gemini 4 Carbon.
Intelligence Index, top 15
Composite of agentic, coding, science and knowledge evals. Higher is better. Hover, tap or tab to a bar to inspect the model.
- 1Gemini 4 Carbon (high)59.4
- 2Claude Opus 5.5 (max with fallback)58.4
- 3Gemini 4 Carbon (medium)57.7
- 4Claude Opus 5.5 (xhigh with fallback)57.0
- 5GPT-6 Astra (max)56.9
- 6GPT-6 Astra (xhigh)55.3
- 7GPT-6 Astra (high)54.9
- 8GPT-6.1 Sol (max)54.5
- 9Claude Sonnet 5.5 (max with fallback)54.2
- 10GPT-6.1 Sol (xhigh)52.3
- 11Claude Opus 5.5 (high with fallback)51.7
- 12GPT-6 Astra (medium)51.5
- 13GPT-6.1 Terra (max)51.0
- 14Gemini 4 Argon (high)50.4
- 15GPT-6.1 Sol (high)50.2
Intelligence vs cost to run the index
USD to run the full Intelligence Index, log scale. Up and left is better.
Glowing line: the best score you can get at each cost. 82 scored models plotted.
Intelligence vs output speed
Median output tokens per second, log scale. Up and right is better.
Glowing line: the best score you can get at each speed. 79 scored models plotted.
Click a column to sort. Prices are USD per 1M tokens; latency is time to first token. Click a row to inspect it.
| Gemini 4 Carbon (high)Google | 59.4 | $7.76 | $4.00 | $20.00 | 95 | 137.6 s | 1M | 2026-10-28 | Closed |
|---|---|---|---|---|---|---|---|---|---|
| Claude Opus 5.5 (max with fallback)Anthropic | 58.4 | $10.04 | $4.00 | $20.00 | 96 | 728.7 s | 1M | 2026-09-22 | Closed |
| Gemini 4 Carbon (medium)Google | 57.7 | $6.59 | $4.00 | $20.00 | – | – | 1M | 2026-10-28 | Closed |
| Claude Opus 5.5 (xhigh with fallback)Anthropic | 57.0 | $6.05 | $4.00 | $20.00 | 82 | 128.8 s | 1M | 2026-09-22 | Closed |
| GPT-6 Astra (max)OpenAI | 56.9 | $6.24 | $10.00 | $50.00 | 72 | 367.6 s | 1M | 2026-09-03 | Closed |
| GPT-6 Astra (xhigh)OpenAI | 55.3 | $3.66 | $10.00 | $50.00 | 68 | 197.4 s | 1M | 2026-09-03 | Closed |
| GPT-6 Astra (high)OpenAI | 54.9 | $3.24 | $10.00 | $50.00 | 62 | 95.7 s | 1M | 2026-09-03 | Closed |
| GPT-6.1 Sol (max)OpenAI | 54.5 | $1.27 | $2.00 | $10.00 | 87 | 331.4 s | 1M | 2026-09-29 | Closed |
| Claude Sonnet 5.5 (max with fallback)Anthropic | 54.2 | $10.02 | $2.00 | $10.00 | 137 | 514.9 s | 1M | 2026-09-28 | Closed |
| GPT-6.1 Sol (xhigh)OpenAI | 52.3 | $0.76 | $2.00 | $10.00 | 77 | 144.1 s | 1M | 2026-09-29 | Closed |
| Claude Opus 5.5 (high with fallback)Anthropic | 51.7 | $3.46 | $4.00 | $20.00 | 80 | 47.5 s | 1M | 2026-09-22 | Closed |
| GPT-6 Astra (medium)OpenAI | 51.5 | $2.73 | $10.00 | $50.00 | 57 | 5.09 s | 1M | 2026-09-03 | Closed |
| GPT-6.1 Terra (max)OpenAI | 51.0 | $0.96 | $1.00 | $5.00 | 107 | 264.3 s | 1M | 2026-10-26 | Closed |
| Gemini 4 Argon (high)Google | 50.4 | $7.68 | $4.00 | $20.00 | 83 | 40.5 s | 1M | 2026-09-30 | Closed |
| GPT-6.1 Sol (high)OpenAI | 50.2 | $0.59 | $2.00 | $10.00 | 66 | 69.6 s | 1M | 2026-09-29 | Closed |
| Muse Wonder Full (max)Meta | 49.7 | – | – | – | – | – | 1M | 2026-10-12 | Closed |
| Claude Fable 5.1 (xhigh with fallback)Anthropic | 49.7 | $10.85 | $10.00 | $50.00 | 62 | 73.5 s | 1M | 2026-09-01 | Closed |
| Claude Sonnet 5.5 (xhigh with fallback)Anthropic | 49.6 | $4.09 | $2.00 | $10.00 | 108 | 32.9 s | 1M | 2026-09-28 | Closed |
| Claude Fable 5.1 (max with fallback)Anthropic | 49.0 | $12.99 | $10.00 | $50.00 | 70 | 263.5 s | 1M | 2026-09-01 | Closed |
| Claude Opus 5.5 (medium with fallback)Anthropic | 48.1 | $2.45 | $4.00 | $20.00 | 80 | 19.9 s | 1M | 2026-09-22 | Closed |
| Claude Fable 5.1 (high with fallback)Anthropic | 47.0 | $7.54 | $10.00 | $50.00 | 58 | 17.1 s | 1M | 2026-09-01 | Closed |
| GPT-6.1 Sol (medium)OpenAI | 46.1 | $0.36 | $2.00 | $10.00 | 63 | 8.14 s | 1M | 2026-09-29 | Closed |
| Kimi K3.1 ProKimi | 46.0 | $3.68 | $2.00 | $10.00 | 31 | 68.3 s | 1M | 2026-10-23 | Closed |
| GPT-6.1 Terra (xhigh)OpenAI | 45.0 | $0.40 | $1.00 | $5.00 | 101 | 68.8 s | 1M | 2026-10-26 | Closed |
| Muse Wonder (max)Meta | 43.3 | $6.95 | $4.00 | $20.00 | 192 | 36.8 s | 1M | 2026-10-12 | Closed |
| GPT-6 Astra (low)OpenAI | 43.0 | $1.42 | $10.00 | $50.00 | 57 | 2.55 s | 1M | 2026-09-03 | Closed |
| Claude Fable 5.1 (medium with fallback)Anthropic | 43.0 | $5.56 | $10.00 | $50.00 | 58 | 12.9 s | 1M | 2026-09-01 | Closed |
| GPT-6.1 Terra (high)OpenAI | 42.3 | $0.26 | $1.00 | $5.00 | 96 | 20.7 s | 1M | 2026-10-26 | Closed |
| Claude Sonnet 5.5 (high with fallback)Anthropic | 41.1 | $1.64 | $2.00 | $10.00 | 107 | 12.6 s | 1M | 2026-09-28 | Closed |
| Claude Fable 5.1 (low with fallback)Anthropic | 38.5 | $4.47 | $10.00 | $50.00 | 56 | 2.78 s | 1M | 2026-09-01 | Closed |
| GPT-6.1 Terra (medium)OpenAI | 38.1 | $0.17 | $1.00 | $5.00 | 91 | 4.90 s | 1M | 2026-10-26 | Closed |
| Claude Haiku 5.5 (max)Anthropic | 36.2 | $0.41 | $0.10 | $0.50 | 235 | 370.1 s | 1M | 2026-10-07 | Closed |
| Grok 4.7 (high)SpaceXAI | 35.0 | $6.18 | $2.00 | $6.00 | 76 | 43.2 s | 488K | 2026-09-21 | Closed |
| Claude Opus 5.5 (low with fallback)Anthropic | 34.5 | $0.98 | $4.00 | $20.00 | 79 | 10.3 s | 1M | 2026-09-22 | Closed |
| GPT-6.1 Sol (low)OpenAI | 34.5 | $0.16 | $2.00 | $10.00 | 61 | 2.48 s | 1M | 2026-09-29 | Closed |
| Claude Haiku 5.5 (xhigh)Anthropic | 34.3 | $0.25 | $0.10 | $0.50 | 200 | 71.5 s | 1M | 2026-10-07 | Closed |
| Grok 4.7 (xhigh)SpaceXAI | 34.3 | $7.51 | $2.00 | $6.00 | 81 | 102.0 s | 488K | 2026-09-21 | Closed |
| Qwen3.8 Max (0902)Alibaba | 33.5 | $10.02 | $2.00 | $6.00 | 40 | 53.5 s | 961K | 2026-09-02 | Closed |
| DeepSeek V4.1 Flash 1017 (max)DeepSeek | 32.5 | $0.60 | $0.30 | $1.20 | 171 | 13.0 s | 1M | 2026-10-17 | Open |
| GLM-5.3 (max)Z AI | 31.6 | $5.01 | $1.40 | $4.40 | 89 | 23.9 s | 1M | 2026-08-18 | Open |
| Grok 4.7 (low)SpaceXAI | 30.9 | $2.92 | $2.00 | $6.00 | 70 | 6.11 s | 488K | 2026-09-21 | Closed |
| Gemini 3.8 Flash (high)Google | 30.6 | $7.77 | $0.75 | $3.75 | 131 | 24.9 s | 1M | 2026-09-02 | Closed |
| GPT-6.1 Terra (non-reasoning)OpenAI | 30.4 | $0.23 | $1.00 | $5.00 | 91 | 0.96 s | 1M | 2026-10-26 | Closed |
| Claude Sonnet 5.5 (medium with fallback)Anthropic | 30.3 | $0.78 | $2.00 | $10.00 | 104 | 1.22 s | 1M | 2026-09-28 | Closed |
| Gemini 3.8 Flash (medium)Google | 30.0 | $6.26 | $0.75 | $3.75 | – | – | 1M | 2026-09-02 | Closed |
| MiMo-V2.6-ProXiaomi | 29.9 | $0.26 | $0.43 | $0.87 | 44 | 49.3 s | 1M | 2026-09-21 | Open |
| Claude Haiku 5.5 (high)Anthropic | 29.6 | $0.16 | $0.10 | $0.50 | 181 | 31.7 s | 1M | 2026-10-07 | Closed |
| Ling 3.1 Flash 1019InclusionAI | 29.5 | $3.19 | $0.30 | $0.90 | – | – | 1M | 2026-10-22 | Open |
| GPT-6.1 Terra (low)OpenAI | 29.2 | $0.071 | $1.00 | $5.00 | 91 | 1.43 s | 1M | 2026-10-26 | Closed |
| MiniMax-M3.1-FlashMiniMax | 28.9 | $0.49 | $0.15 | $0.60 | 136 | 15.6 s | 1M | 2026-10-19 | Open |
| DeepSeek V4.1 Flash 1017 (high)DeepSeek | 28.5 | – | $0.30 | $1.20 | – | – | 1M | 2026-10-17 | Open |
| Step 5 PreviewStepFun | 28.5 | $2.35 | $1.00 | $2.70 | 146 | 14.6 s | 1M | 2026-09-18 | Open |
| GPT-6 Luna (max)OpenAI | 27.8 | $0.15 | $0.10 | $0.50 | 117 | 112.8 s | 1M | 2026-09-22 | Closed |
| GLM-5.3-FlashZ AI | 27.8 | $0.58 | $0.15 | $0.50 | 76 | 27.8 s | 1M | 2026-08-26 | Open |
| MiMo-V2.6-FlashXiaomi | 26.7 | $0.13 | $0.14 | $0.28 | 37 | 58.1 s | 1M | 2026-09-21 | Open |
| Mistral Large 4 (high)Mistral | 26.0 | $2.06 | $0.70 | $2.10 | – | – | 1M | 2026-10-29 | Open |
| Claude Sonnet 5.5 (low with fallback)Anthropic | 25.3 | $0.51 | $2.00 | $10.00 | 106 | 1.15 s | 1M | 2026-09-28 | Closed |
| Muse Spark Open (xhigh)Meta | 24.7 | $2.02 | $1.25 | $4.25 | 193 | 11.1 s | 1M | 2026-10-15 | Open |
| Claude Haiku 5.5 (medium)Anthropic | 24.6 | $0.092 | $0.10 | $0.50 | 172 | 13.8 s | 1M | 2026-10-07 | Closed |
| GPT-6 Luna (xhigh)OpenAI | 23.3 | $0.10 | $0.10 | $0.50 | 128 | 40.4 s | 1M | 2026-09-22 | Closed |
| Mistral Large 4 (medium)Mistral | 23.1 | $1.41 | $0.70 | $2.10 | – | – | 1M | 2026-10-29 | Open |
| DeepSeek V4 Pro 0813 (max)DeepSeek | 23.1 | $1.77 | $1.32 | $3.96 | 94 | 23.2 s | 1M | 2026-08-13 | Open |
| GLM-5.3 (low)Z AI | 23.1 | $1.57 | $1.40 | $4.40 | 80 | 26.5 s | 1M | 2026-08-18 | Open |
| Gemini 3.8 Flash (low)Google | 22.4 | – | $0.75 | $3.75 | – | – | 1M | 2026-09-02 | Closed |
| GPT-6 Luna (high)OpenAI | 21.4 | $0.073 | $0.10 | $0.50 | 119 | 29.0 s | 1M | 2026-09-22 | Closed |
| Nanbeige4.3-33BNanbeige | 20.4 | – | $0 | $0 | 160 | 14.0 s | 512K | 2026-10-25 | Open |
| Qwen3.8 27B (xhigh)Alibaba | 20.1 | $3.53 | $0.50 | $3.00 | 41 | 52.7 s | 250K | 2026-08-14 | Open |
| MiniMax-M3MiniMax | 18.7 | $1.29 | $0.30 | $1.20 | 120 | 17.4 s | 1M | 2026-06-01 | Open |
| Claude Haiku 5.5 (low)Anthropic | 18.6 | $0.044 | $0.10 | $0.50 | 167 | 8.83 s | 1M | 2026-10-07 | Closed |
| DeepSeek V4.1 Flash 1017 (non-reasoning)DeepSeek | 18.5 | $0.43 | $0.30 | $1.20 | 185 | 1.38 s | 1M | 2026-10-17 | Open |
| GPT-6 Luna (medium)OpenAI | 18.3 | $0.042 | $0.10 | $0.50 | – | – | 1M | 2026-09-22 | Closed |
| K2 Horizon 375B A23BInstitute of Foundation Models | 17.8 | <$0.01 | $0 | $0 | 720 | 5.15 s | 512K | 2026-09-03 | Open |
| Beam 1.0Reflection | 17.7 | $0.28 | $0.40 | $1.20 | 361 | 6.16 s | 1M | 2026-10-16 | Open |
| Qwen3.8 27B (medium)Alibaba | 17.5 | $2.16 | $0.50 | $3.00 | 51 | 43.2 s | 250K | 2026-08-14 | Open |
| Qwen3.8 27B (low)Alibaba | 15.7 | $1.91 | $0.50 | $3.00 | 48 | 45.4 s | 250K | 2026-08-14 | Open |
| Nemotron 3.5 Super VLNVIDIA | 13.6 | $0.54 | $0.20 | $0.50 | 479 | 4.89 s | 1M | 2026-10-15 | Open |
| Inkling SmallThinking Machines | 12.9 | $0.20 | $0.30 | $1.20 | 230 | 10.3 s | 1M | 2026-07-30 | Open |
| Inkling (xhigh)Thinking Machines | 12.5 | $1.35 | $1.00 | $4.05 | 211 | 11.2 s | 1M | 2026-07-15 | Open |
| Nemotron 3 UltraNVIDIA | 11.8 | $2.23 | $0.60 | $2.70 | 161 | 15.3 s | 256K | 2026-06-04 | Open |
| Gemini 3.5 Flash-LiteGoogle | 11.7 | $0.69 | $0.30 | $2.50 | 423 | 6.22 s | 1M | 2026-07-21 | Closed |
| Qwen3.8 27B (non-reasoning)Alibaba | 10.0 | $3.77 | $0.50 | $3.00 | 46 | 3.96 s | 250K | 2026-08-14 | Open |
| GPT-6 Luna (low)OpenAI | 9.7 | $0.005 | $0.10 | $0.50 | 121 | 2.81 s | 1M | 2026-09-22 | Closed |
| GPT-6 Luna (non-reasoning)OpenAI | 9.4 | $0.071 | $0.10 | $0.50 | 112 | 0.81 s | 1M | 2026-09-22 | Closed |
| Muse Glimmer (high)Meta | 6.3 | $0.16 | $0.30 | $1.20 | 120 | 17.8 s | 128K | 2026-08-10 | Open |
| Command A+Cohere | 6.2 | <$0.01 | $0 | $0 | 154 | 13.5 s | 188K | 2026-05-20 | Open |
| Nemotron 3.5 LightningNVIDIA | 5.5 | $0.29 | $0.060 | $0.20 | 295 | 7.35 s | 1M | 2026-08-11 | Open |
| Granite 4.2 8BIBM | 5.1 | $0.041 | $0.060 | $0.25 | 61 | 33.7 s | 128K | 2026-08-25 | Open |
| Granite 4.2 3BIBM | 4.7 | – | – | – | – | – | 128K | 2026-08-25 | Open |
| Mistral Small 4Mistral | 4.3 | $0.040 | $0.15 | $0.60 | 158 | 13.4 s | 250K | 2026-03-16 | Open |
| Nemotron 3 Nano 4BNVIDIA | 1.8 | – | – | – | – | – | 256K | 2026-03-16 | Open |
Which lab has the best LLM? The top model from each lab
Each lab's highest-scoring model and how many of its models are scored. Links go to the lab's own site.
- Google7 of 21 scoredGemini 4 Carbon (high)59.4 · $7.76 per index runGoogle DeepMind
- Anthropic20 of 20 scoredClaude Opus 5.5 (max with fallback)58.4 · $10.04 per index runAnthropic
- OpenAI22 of 31 scoredGPT-6 Astra (max)56.9 · $6.24 per index runOpenAI
- Meta4 of 10 scoredMuse Wonder Full (max)49.7 · no cost figureAI at Meta
- Kimi1 of 3 scoredKimi K3.1 Pro46.0 · $3.68 per index runMoonshot AI
- SpaceXAI3 of 3 scoredGrok 4.7 (high)35.0 · $6.18 per index runxAI
- Alibaba5 of 29 scoredQwen3.8 Max (0902)33.5 · $10.02 per index runQwen (Alibaba)
- DeepSeek4 of 6 scoredDeepSeek V4.1 Flash 1017 (max)32.5 · $0.60 per index runDeepSeek
- Z AI3 of 3 scoredGLM-5.3 (max)31.6 · $5.01 per index runZ.ai
- Xiaomi2 of 2 scoredMiMo-V2.6-Pro29.9 · $0.26 per index runXiaomi MiMo
- InclusionAI1 of 7 scoredLing 3.1 Flash 101929.5 · $3.19 per index runNo official site confirmed
- MiniMax2 of 2 scoredMiniMax-M3.1-Flash28.9 · $0.49 per index runMiniMax
- StepFun1 of 2 scoredStep 5 Preview28.5 · $2.35 per index runStepFun
- Mistral3 of 10 scoredMistral Large 4 (high)26.0 · $2.06 per index runMistral AI
- Nanbeige1 of 2 scoredNanbeige4.3-33B20.4 · no cost figureNo official site confirmed
- Institute of Foundation Models1 of 9 scoredK2 Horizon 375B A23B17.8 · <$0.01 per index runNo official site confirmed
- Reflection1 of 1 scoredBeam 1.017.7 · $0.28 per index runReflection AI
- NVIDIA4 of 15 scoredNemotron 3.5 Super VL13.6 · $0.54 per index runNVIDIA
- Thinking Machines2 of 2 scoredInkling Small12.9 · $0.20 per index runThinking Machines Lab
- Cohere1 of 4 scoredCommand A+6.2 · <$0.01 per index runCohere
- IBM2 of 8 scoredGranite 4.2 8B5.1 · $0.041 per index runIBM Granite
How do you read an LLM leaderboard? What each column means
- Intelligence Index
- A composite of agentic, coding, science and knowledge evaluations. Higher is better. It is a single number for ranking, not a promise about your task.
- Index cost
- US dollars to run the full Intelligence Index with that model. A rough measure of how expensive the model is to use for hard work, not a price list.
- Input and output price
- US dollars per million tokens. Reasoning models often spend many extra output tokens thinking, so the cost of a finished job can differ from the price per token.
- Tokens per second
- Median output speed on the lab's own API. Faster models feel better in chat and agents, but speed rarely comes with the top score.
- Latency
- Time to the first answer token, in seconds. Reasoning models can take longer because they think before answering.
- Context
- The longest input the model accepts, in tokens. Long context does not guarantee good long-context reasoning; see the Long context benchmark.
- Open weights
- The model files are published so you can run them yourself. Licences differ, so read the licence before commercial use.
- Benchmarks
- Percentage scores on individual evaluations such as Terminal-Bench 4, FrontierCode 2, SciCode, GPQA and HLE. Hover a benchmark in the detail card for what it measures.
LLM leaderboard questions, answered
What is the best LLM in 2026?
On our leaderboard, Gemini 4 Carbon (high) from Google scores highest at 59.4 on the Intelligence Index. GPT-6.1 Sol (max) comes within 4.9 points at $1.27 per index run, against $7.76 for the leader (data as of 2026-11-02).
Which LLM is the best value near the top?
GPT-6.1 Sol (max) from OpenAI: it scores 54.5 and costs $1.27 to run the full index, 84% less than the leader. We count only models within 5 points of the top score that have a cost figure.
What is the fastest LLM?
Among scored models, K2 Horizon 375B A23B from Institute of Foundation Models is fastest at 720 output tokens per second, but it scores 17.8, 41.7 points below the leader. Speed is the median on the lab's own API.
What is the best open-weights LLM?
DeepSeek V4.1 Flash 1017 (max) from DeepSeek scores highest among open-weights models at 32.5, 26.9 points below the leader. 34 of the 90 scored models have published weights; check each model's licence before commercial use.
Why do some models have no score?
We track 278 models but only 90 have a current Intelligence Index score. The rest are older or smaller models we list for price, speed and benchmark context; use the "Scored only" switch to include them. Cost and speed figures are also missing for some scored models.
Sources and limits
This leaderboard was compiled by the TheBestSuperintelligence editorial team from published benchmark and pricing sources and is dated 2026-11-02. We have not itemised the source of every column. Intelligence scores follow the Intelligence Index v5.0 published by Artificial Analysis, and for the 25 models this leaderboard shares with our Frontier Gap Report the scores agree after rounding. Only 90 of the 278 models have an index score, and cost and speed figures are missing for some of those, so a blank means "not available", not zero. Prices, speeds and costs change often and some are vendor-reported; check the lab's page before you buy. "Open weights" says the weights are published, not that the licence allows every use. Lab links were opened on 2026-11-02; labs without a confirmed official site are not linked.
Keep reading: the Frontier Gap Report, how to compare frontier AI models, the top 10 AI companies and the best AI agents for work.