Rank 270+ AI models by the job
Pick a scenario, drag the weights, set a budget. Free, ad-free and built on open data.
Models ranked
267
Benchmarks
38
Scenarios
14
Last refresh
16 Nov 2026
Leaderboard
Each scenario weights benchmarks differently. Switch between them and the leaderboard re-ranks live.
A pair-programming assistant for IDE / agent loops. Heavy on coding benchmarks, with a real-world agentic component (SWE-bench) and some weight on cost since coding loops burn tokens. A small slice goes to the recovery-rate reliability axis because real-world agent loops live or die on whether the model self-corrects.Scores here rest on a median 25% of this scenario's benchmark weight; 10 of 185 models have 80% or more. See the evidence map.
Showing the top 50 of 76 models, sorted by scenario score, descending.
| Shortlist and details | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Google vision | 80.450% cov. | $20 | 1.0M | 149 | 21.5 | — | — | — | — | — | ||||
| 2 | Google vision | 80.130% cov. | $69 | 1.0M | 44.5 | 4.30 | $4.31 | 64.2 | — | — | — | ||||
| 3 | Google vision | 78.750% cov. | $78 | 2.0M | 39.0 | 4.47 | — | — | — | — | — | ||||
| 4 | Google vision | 78.030% cov. | $26 | 1.0M | 110 | 4.36 | $6.26 | 19.0 | — | — | — | ||||
| 5 | MiniMax | 76.740% cov. | $18 | 198k | 152 | 4.90 | — | — | — | — | — | — | |||
| 6 | Anthropic vision | 76.630% cov. | $276 | 1.0M | 11.3 | 4.27 | $2.45 | 81.1 | — | — | — | ||||
| 7 | Zhipu AI (GLM) | 75.640% cov. | $34 | 200k | 80.4 | 4.18 | — | — | — | — | — | — | |||
| 8 | DeepSeek | 74.970% cov. | $31 | 128k | 86.8 | 4.36 | — | — | — | — | |||||
| 9 | OpenAI vision | 74.630% cov. | $8 | 1.1M | 325 | 8.51 | $0.33 | 57.8 | — | — | — | ||||
| 10 | OpenAI vision | 74.330% cov. | $69 | 1.1M | 40.9 | 5.44 | $0.36 | 35.7 | — | — | — | ||||
| 11 | DeepSeek | 74.230% cov. | $4 | 1.0M | 619 | 13.0 | — | 37.5 | — | — | — | ||||
| 12 | Anthropic vision | 73.530% cov. | $138 | 1.0M | 20.9 | 10.1 | $0.78 | 39.1 | — | — | — | ||||
| 13 | DeepSeek | 73.455% cov. | $17 | 128k | 149 | 5.16 | — | — | — | — | — | ||||
| 14 | OpenAI vision | 73.365% cov. | $120 | 400k | 24.1 | 2.64 | — | — | — | — | |||||
| 15 | OpenAI vision | 72.830% cov. | $3 | 1.1M | 675 | 5.77 | $0.04 | 35.7 | — | — | — | ||||
| 16 | OpenAI vision | 72.750% cov. | $84 | 400k | 33.5 | 4.52 | — | — | — | — | — | ||||
| 17 | Anthropic vision | 72.655% cov. | $345 | 200k | 8.62 | 6.94 | — | 19.7 | — | — | |||||
| 18 | Google vision | 72.4 | $60 | 2.0M | 45.8 | 3.76 | — | — | — | ||||||
| 19 | Zhipu AI (GLM) | 72.450% cov. | $53 | 205k | 50.5 | 7.42 | — | — | — | — | — | ||||
| 20 | Alibaba (Qwen) | 72.440% cov. | $16 | 262k | 154 | 5.75 | — | — | — | — | — | — | |||
| 21 | Alibaba (Qwen) | 72.070% cov. | $11 | 131k | 233 | 4.49 | — | — | — | — | |||||
| 22 | Zhipu AI (GLM) | 71.940% cov. | $34 | 131k | 75.8 | 4.19 | — | — | — | — | — | — | |||
| 23 | OpenAI vision | 71.830% cov. | $78 | 1.1M | 35.2 | 8.82 | $3.73 | 33.2 | — | — | — | ||||
| 24 | Anthropic vision | 71.850% cov. | $345 | 200k | 8.52 | 4.26 | — | — | — | — | — | ||||
| 25 | OpenAI | 71.7 | $66 | 200k | 40.7 | 3.63 | — | — | — | ||||||
| 26 | Moonshot (Kimi) | 71.265% cov. | $34 | 256k | 74.8 | 11.1 | — | — | — | — | |||||
| 27 | Anthropic vision | 71.030% cov. | $690 | 1.0M | 4.36 | 6.04 | $5.56 | 30.4 | — | — | — | ||||
| 28 | OpenAI vision | 70.330% cov. | $195 | 400k | 14.4 | 4.89 | — | 38.6 | — | — | — | ||||
| 29 | OpenAI | 70.2 | $120 | 200k | 22.5 | 7.35 | — | — | — | ||||||
| 30 | OpenAI vision | 69.830% cov. | $138 | 1.1M | 19.7 | 6.26 | — | 39.3 | — | — | — | ||||
| 31 | Zhipu AI (GLM) | 69.530% cov. | $73 | 1.0M | 35.3 | 7.52 | $1.57 | 17.7 | — | — | — | ||||
| 32 | OpenAI vision | 69.430% cov. | $345 | 1.1M | 8.23 | 4.29 | $2.73 | 34.5 | — | — | — | ||||
| 33 | OpenAI vision | 69.250% cov. | $60 | 400k | 43.5 | 1.71 | — | — | — | — | — | ||||
| 34 | Anthropic vision | 68.950% cov. | $207 | 200k | 13.2 | 3.31 | — | — | — | — | — | ||||
| 35 | Anthropic vision | 68.430% cov. | $345 | 1.0M | 8.10 | 6.89 | $11.82 | 13.4 | — | — | — | ||||
| 36 | Anthropic vision | 68.130% cov. | $345 | 200k | 8.06 | 3.50 | — | 38.5 | — | — | — | ||||
| 37 | Moonshot (Kimi) vision | 66.930% cov. | $138 | 1.0M | 18.8 | 3.60 | $3.68 | 20.7 | — | — | — | ||||
| 38 | OpenAI vision | 66.730% cov. | $69 | 1.1M | 36.1 | 4.99 | $2.47 | 37.6 | — | — | — | ||||
| 39 | OpenAI vision | 65.150% cov. | $12 | 400k | 183 | 3.52 | — | — | — | — | — | ||||
| 40 | Anthropic vision | 63.750% cov. | $69 | 200k | 34.3 | 2.21 | $0.53 | — | — | — | — | ||||
| 41 | Anthropic vision | 62.430% cov. | $138 | 1.0M | 17.4 | 9.35 | — | 16.7 | — | — | — | ||||
| 42 | Google vision | 62.230% cov. | $15 | 1.0M | 142 | 9.40 | $0.69 | 0.0 | — | — | — | ||||
| 43 | DeepSeek | 61.7 | $17 | 128k | 120 | 12.3 | — | — | — | ||||||
| 44 | xAI | 61.470% cov. | — | 1.0M | — | — | — | — | — | — | |||||
| 45 | OpenAI vision | 61.240% cov. | $98 | 400k | 23.8 | 6.00 | — | — | — | — | — | ||||
| 46 | Anthropic vision | 60.930% cov. | $345 | 1.0M | 7.16 | 1.66 | — | 9.2 | — | — | — | ||||
| 47 | Anthropic vision | 60.930% cov. | $207 | 200k | 11.6 | 6.62 | — | 20.1 | — | — | — | ||||
| 48 | xAI vision | 60.840% cov. | $345 | 256k | 7.15 | 3.23 | — | — | — | — | — | ||||
| 49 | Google vision | 59.6 | $15 | 1.0M | 134 | 4.48 | — | — | — | ||||||
| 50 | OpenAI | 57.4 | $66 | 200k | 31.4 | 5.66 | — | — | — |
- Scenario score
- weighted benchmark average, 0–100; blended with cost when the scenario weights cost
- Score/$
- score per blended $/M tokens (3 in : 1 out)
- Score/s
- score per second for a 500-token answer
- Stability
- consistency of repeated answers, from an independent monitor
- % cov.
- share of the benchmark weight with data
- high contamination risk
- saturated benchmark
- think
- reasoning-effort variant of the same model
- More
- methodology
Showing 76 of 267 models. Tap or hover any score for the raw value, its date and contamination risk. Click a model for its full profile.
Cost vs quality
Models on the Pareto frontier (highlighted) give you the best quality at their cost tier.
Shortlists
Highest Analyzer Index; effort variants of the same model count once.