AI model leaderboard · updated Oct 20, 2026
Compare every frontier and open-weights AI model
Benchmark scores with uncertainty intervals, live prices and context windows, all in one place, with every number traceable to its source. Our own speed measurements are coming soon.
Leaders right now
Top 10 models
Overall score with 90% interval. Overlapping intervals mean the order is not certain.
| # | Model | Score | 90% interval | Price |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 AnthropicProprietary | 86.7 | $4.00 / $20.00 | |
| 2 | GPT-6.1 Sol OpenAIProprietary | 84.2 | $2.00 / $10.00 | |
| 3 | GPT-6 Astra OpenAIProprietary | 84.0 | $10.00 / $50.00 | |
| 4 | Claude Fable 5.1 AnthropicProprietary | 83.2 | $10.00 / $50.00 | |
| 5 | Gemini 4 Argon GoogleProprietary | 82.9 | $2.00 / $10.00 | |
| 6 | Claude Sonnet 5.5 AnthropicProprietary | 82.2 | $2.00 / $10.00 | |
| 7 | Claude Fable 5 AnthropicProprietary | 81.3 | $10.00 / $50.00 | |
| 8 | Claude Opus 5 AnthropicProprietary | 80.1 | $5.00 / $25.00 | |
| 9 | GPT-6 Sol OpenAIProprietary | 78.6 | $2.00 / $10.00 | |
| 10 | GPT-5.6 Sol OpenAIProprietary | 75.3 | $4.00 / $20.00 |
Best model for each use case
Picked mechanically from the data below.
- Highest quality, regardless of priceClaude Opus 5.5Score 86.7Verified
- Best open weights (self-host or fine-tune)MiMo-V2.6-ProHighest-ranked open-weights modelProvisional
- Best value (score per dollar, top half)GLM-5.3-Flash896 points per $ (blended)Provisional
- Cheapest model in the top halfGLM-5.3-Flash$0.065 / 1M blendedProvisional
- Long documents and codebasesLlama 4 ScoutLargest context among ranked modelsVerified
Score vs price
Each point is a ranked model. Blended price assumes 3 input tokens per output token. Click a point to open the model.
Open vs proprietary over time
Best score released up to each month.
Recently released
- MiniMax M3.1 FlashMiniMax · Oct 19, 2026Score 67.11M ctx
- Mercury 3Inception · Oct 19, 2026Not ranked yet1M ctx
- DeepSeek V4.1 Flash 1017DeepSeek · Oct 17, 2026Score 65.71M ctx
- GlimpseElorian · Oct 17, 2026Not ranked yet262K ctx
- Beam 1.0 501B-A23BReflection AI · Oct 16, 2026Score 51.61M ctx
- Perplexity Decider Flash v1.2 4BPerplexity · Oct 16, 2026Not ranked yet262K ctx
- Perplexity Decider v1.2 27BPerplexity · Oct 16, 2026Not ranked yet262K ctx
- Muse Spark OpenMeta · Oct 15, 2026Score 62.81M ctx
Frequently asked questions
Answers are generated from the current data.
›What is the best AI model right now?
Claude Opus 5.5 (Anthropic) leads with a score of 86.7, ahead of GPT-6.1 Sol (OpenAI) at 84.2. Their 90% intervals overlap, so treat the top spots as a statistical tie.
›What is the best open-weights model?
MiMo-V2.6-Pro (Xiaomi) is the highest-ranked open-weights model at 73.7, 13.0 points behind the overall leader.
›Which top model is the cheapest?
Among the top 20, MiMo-V2.6-Pro (Xiaomi) has the lowest blended price at $0.54 per million tokens (3:1 input to output).
›Which model is best for coding?
Claude Fable 5.1 (Anthropic) has the highest coding score (91.2).
›Which ranked model has the longest context window?
Llama 4 Scout (Meta) accepts up to 10,000,000 tokens.
›Which model is the fastest?
We have not published speed measurements yet. Speed is only shown when we measured it ourselves.
›How are models ranked?
Overall and category scores come from BenchLM's published rankings, which aggregate benchmark results reported in system cards, papers and public evaluations. Each score has a 90% interval; models whose intervals overlap are not reliably different. See the Sources & credits page.
278 benchmarks · see sources & credits.