← All models

o3

Language Available Full comparison ↗

openai/o3 · by OpenAI · decoder-only-transformer

Pricing — 1 offering(s)

OpenAI

View on OpenAI ↗

About this offering: OpenAI o3 reasoning model. Extended thinking budget for complex multi-step tasks. 200K context window. Includes reasoning tokens billed as output.

Input tokens

  • $10.00 / 1M tokens (input) Historical 2025-04-16 → 2026-08-28
  • $2.00 / 1M tokens (input) Current 2026-08-29 → present

Output tokens

  • $40.00 / 1M tokens (output) Historical 2025-04-16 → 2026-08-28
  • $8.00 / 1M tokens (output) Current 2026-08-29 → present

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

See how o3 fits into a cost-aware routing setup

See how →

Capability profile

How o3 rates across core capability dimensions, with the task-level evidence behind each rating.

reasoning Strong

Designed for extended multi-step reasoning; reasoning tokens billed as output.

coding Strong

SWE-bench Verified (500 human-verified GitHub issues; patch must pass the repo's own tests; no file-location hints): resolves 69.1% — fails roughly 1 issue in 3. Vendor-reported (OpenAI, o3 launch; no custom scaffolding — higher with agent scaffolds).

tool use Strong
instruction following Strong
context window Strong

200K context.

multilingual Moderate
speed Weak

Slower than GPT-4o due to internal reasoning budget.

cost efficiency Weak

Coding benchmarks

View the full coding benchmark leaderboard →
Benchmark Score Harness Source
swe-bench-verified ↗ 69.1% 2025-04 agentless 🏢 vendor ↗
aider-polyglot ↗ 76.9% 2025-04 aider 📊 leaderboard ↗
livecodebench ↗ 65.3% 2025-04 — 📊 leaderboard ↗

Operator guidance

Route here when quality matters more than latency or cost for hard reasoning tasks. Not a replacement for GPT-4o on standard chat workloads — more expensive and slower.

Use cases

Limitations

Citations