Best AI models for one-shot coding
One-shot capability is whether a model can take a prompt and deliver working front end, back end and database without multiple correction rounds. It's the single most important trait in a coding model.
S tier — Elite
The one-shot king — practitioners call it the best model in the world.
- Claude Fable 5.1 — $10 / $50 per 1M tokens · Intelligence 53
A tier — Excellent
Both reliably one-shot features; GPT-5.6 Sol excels with sub-agents.
- GPT-5.6 Sol — $4 / $20 per 1M tokens · Intelligence 47
- Claude Opus 5 — $5 / $25 per 1M tokens · Intelligence 51
B tier — Good
Not quite S-tier, but consistently strong.
- Kimi K3 — $3 / $15 per 1M tokens · Intelligence 44
D tier — Needs help
You'll usually need to work with these models to finish the job.
- Grok 4.6 — $2 / $6 per 1M tokens · Intelligence 44
- Qwen3.8 2.4T A95B — $2 / $6 per 1M tokens · Intelligence 40
E tier — Frustrating
Multiple passes typically required.
- GPT-5.6 Terra — $2 / $12 per 1M tokens · Intelligence 42
- GLM-5.3 (max) — $1.40 / $4.40 per 1M tokens · Intelligence 45
F tier — Multiple shots
Fast or cheap, but weak at one-shot reliability.
- Gemini 3.1 Pro — $2 / $12 per 1M tokens · Intelligence 30
- Gemini 3.7 Flash — $0.75 / $3.75 per 1M tokens · Intelligence 39
- GPT-5.6 Luna — $0.20 / $1.20 per 1M tokens · Intelligence 38
- DeepSeek V4 Pro — $0.66 / $1.98 per 1M tokens · Intelligence 36
- DeepSeek V4.1 Flash — $0.15 / $0.60 per 1M tokens · Intelligence 40
Frequently asked questions
Which AI models are best for best ai models for one-shot coding?
The top tier is currently Claude Fable 5.1. The one-shot king — practitioners call it the best model in the world.
How are these AI model rankings made?
These rankings combine practitioner consensus from the frontier-coding community with our own live pricing data (intelligence index, tokens per dollar). They are qualitative judgments, not benchmarks — always cross-check with the calculator for your own workload.