Best AI model for computer use
The best AI model for computer use on current benchmarks is Claude Opus 5 (Anthropic): as of October 19, 2026 it has the highest OSWorld 2.0 score TensorFeed tracks, 70.6% (vendor-reported), 4.5 points ahead of Claude Fable 5 (66.1%), among 11 models scored. On Terminal-Bench 4.0, Claude Opus 5.5 leads (65.2%). Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.1 and GPT-6 Astra have no OSWorld 2.0 score yet, so they are not ranked here.
Ranked on OSWorld 2.0. Benchmark data as of October 19, 2026; prices as of October 19, 2026.
OSWorld 2.0: 58.3% (vendor-reported). $3.00 / $15.00 per 1M tokens.
Ranking on OSWorld 2.0
Each score links to its source; vendor-reported and independent figures are labeled and are not interchangeable.
| # | Model | Provider | OSWorld 2.0 | Price in / out | TFII | Terminal-Bench 4.0 |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5 | Anthropic | 70.6%vendor | $5.00 / $25.00 | 83.9 | 53.54% |
| 2 | Claude Fable 5 | Anthropic | 66.1% | $10.00 / $50.00 | 81.6 | 41.41% |
| 3 | GPT-5.6 Sol | OpenAI | 62.6%vendor | $4.00 / $20.00 | 74.5 | 37.88% |
| 4 | Kimi K3 | Moonshot AI | 58.3%vendor | $3.00 / $15.00 | 65.1 | 17.17% |
| 5 | Claude Opus 4.8 | Anthropic | 55.7% | $5.00 / $25.00 | 67.9 | 23.23% |
| 6 | Qwen3.8-Flash | Alibaba | 52.3%vendor | $0.15 / $0.47 | - | - |
| 7 | GPT-5.6 Terra | OpenAI | 50.2%vendor | $2.00 / $12.00 | 60.9 | 22.73% |
| 8 | Qwen3.8 27B | Alibaba | 48%vendor | $0.50 / $3.00 | 75.2 2024 set | - |
| 9 | Gemini 3.7 Flash | 47.9%vendor | - | 52.3 | 12.12% | |
| 10 | GPT-5.6 Luna | OpenAI | 45.6%vendor | $0.20 / $1.20 | 53.2 | 11.62% |
| 11 | Gemini 3.6 Flash | 33.8%vendor | $0.75 / $3.75 | 78.1 2024 set | - |
Prices are list prices per 1M input / output tokens. TFII is the headline TensorFeed Intelligence Index score; scores marked 2024 set come from older benchmarks and are not comparable with the rest.
Not ranked here yet
These models lead the TensorFeed Intelligence Index but have no published OSWorld 2.0 score, so they cannot be placed in this ranking. A missing score is not a low one.
- Claude Opus 5.5 (Anthropic): TFII #1, 97.1
- Claude Sonnet 5.5 (Anthropic): TFII #2, 93.7
- Claude Fable 5.1 (Anthropic): TFII #3, 89.9
- GPT-6 Astra (OpenAI): TFII #5, 83.3
How this is ranked
OSWorld 2.0 hands the model a mouse and a real desktop: 108 workflows across seven professional domains, a median of 1.6 hours for a skilled human and hundreds of tool calls each, graded against weighted checkpoints. It is the one column that says whether an agent can drive software a person normally clicks through.
Most OSWorld 2.0 figures are vendor-published partial scores at a 500-step budget. The benchmark authors headline binary completion, which can run more than thirty points lower on the same run, so do not mix these numbers with ones quoted elsewhere.
The full board, scoring notes and score ranges are on the OSWorld 2.0 leaderboard. The same scores are free as JSON at /api/benchmarks.
Questions
What is the best AI model for computer use?
The best AI model for computer use on current benchmarks is Claude Opus 5 (Anthropic): as of October 19, 2026 it has the highest OSWorld 2.0 score TensorFeed tracks, 70.6% (vendor-reported), 4.5 points ahead of Claude Fable 5 (66.1%), among 11 models scored. On Terminal-Bench 4.0, Claude Opus 5.5 leads (65.2%). Claude Opus 5.5, Claude Sonnet 5.5, Claude Fable 5.1 and GPT-6 Astra have no OSWorld 2.0 score yet, so they are not ranked here.
Which of the top AI models for computer use is cheapest?
Of the five highest OSWorld 2.0 scores TensorFeed tracks, the lowest-priced model with a list price is Kimi K3 (58.3%, vendor-reported), at $3.00 input and $15.00 output per 1M tokens as of October 19, 2026.
Why rank AI models for computer use on OSWorld 2.0?
OSWorld 2.0 hands the model a mouse and a real desktop: 108 workflows across seven professional domains, a median of 1.6 hours for a skilled human and hundreds of tool calls each, graded against weighted checkpoints. It is the one column that says whether an agent can drive software a person normally clicks through.
Why is Claude Opus 5.5 not ranked for computer use?
Claude Opus 5.5 has no published OSWorld 2.0 score in TensorFeed's data yet, so it cannot be ranked here; a missing score is not a low one. It is 1st on the TensorFeed Intelligence Index's 2026 benchmark set (as of October 19, 2026), and it appears on this page once a OSWorld 2.0 score is published.