✦ Our pick · Algorithms, data & scientific code
GPT-6.1 Sol at medium effort, $23.41 per finished task, is the cheapest you can get today. Within our measurement error it ties with Muse Spark 1.1, Claude Opus 5.5, Claude Sonnet 5.5 and GPT-6 Sol: equally good bets, worth trying two.
Updated Oct 13, 2026, every 3 hours. See it live with your plans and costs.
The top 10 for algorithms, data & scientific code
Numerical code, data processing, algorithms. Cost per finished task: the tokens at each model's best effort level for this work, plus your time fixing failed attempts ($93 an hour). Success is the share of such tasks a model gets right the first time, on the benchmarks below.
| # | Model | Effort | Success | $/task |
|---|---|---|---|---|
| 1 | Gemini 4 Argon Google | high | 61.2% partly measured | $22.41 |
| 2 | GPT-6.1 Sol OpenAI | medium | 58.7% partly measured | $23.41 |
| 3 | Muse Spark 1.1 Meta | xhigh | 58.8% partly measured | $23.74 |
| 4 | Claude Opus 5.5 Anthropic | medium | 60.0% measured | $23.92 |
| 5 | Claude Sonnet 5.5 Anthropic | xhigh | 58.6% measured | $25.21 |
| 6 | GPT-6 Sol OpenAI | high | 57.8% measured | $25.57 |
| 7 | MiMo-V2.6-Pro Xiaomi | default | 54.7% measured | $25.99 |
| 8 | GPT-5.6 Sol OpenAI | medium | 58.2% measured | $26.05 |
| 9 | GLM-5.3 Z AI | max | 55.6% measured | $26.12 |
| 10 | GPT-6 Luna OpenAI | max | 54.5% measured | $26.24 |
Best within your tool
| Tool | Model | Effort | $/task |
|---|---|---|---|
| Cursor | Claude Opus 5.5 | medium | $23.92 |
| Claude Code | Claude Opus 5.5 | medium | $23.92 |
| Codex | GPT-6.1 Sol | medium | $23.41 |
| Antigravity | Claude Opus 5.5 | medium | $23.92 |
| GitHub Copilot | GPT-6.1 Sol | medium | $23.41 |
If tokens come with your plan
Tokens included in a plan cost you nothing extra, so the pick is the most reliable model in it, at the leanest effort level within our error of the best (the top levels only use up the plan sooner).
| Plan | Model | Effort |
|---|---|---|
| Claude Pro | Claude Opus 5.5 | high |
| Claude Max | Claude Opus 5.5 | high |
| ChatGPT Plus or Business | GPT-6.1 Sol | high |
| ChatGPT Pro | GPT-6.1 Sol | high |
| Google AI Pro | Claude Opus 5.5 | high |
| Google AI Ultra | Claude Opus 5.5 | high |
Measured on
- SciCode · Artificial Analysis (SciCode authors). 288 subproblems from 80 real laboratory problems across 16 scientific disciplines. Measured for 39 of the 42 ranked models.
- IOI · Vals AI. International Olympiad in Informatics problems: hard algorithms under tests. Measured for 30 of the 42 ranked models, results as of Oct 12, 2026.
- ALE-Bench · Sakana AI (via Epoch AI). AtCoder Heuristic Contest problems: long optimisation tasks, scored as a contestant performance rating. Measured for 26 of the 42 ranked models, results as of Oct 13, 2026.
- Terminal-Bench Science · Vals AI. Research workflows contributed by practising scientists in five fields, done in a real terminal. Measured for 31 of the 42 ranked models, results as of Oct 12, 2026.
How every number is worked out, with the errors of our estimates: the method.