Fast o-series model for compact reasoning, coding, and tool use
Specification
openai listing
Context
200K
Max output
100K
Released
2025-04-16
Knowledge cutoff
2024-05
Retires
2026-10-23
Open weights
no
Input
text, image
Output
text
Price
US dollars per million tokens · openai list price
Input
$1.10
Output
$4.40
Cache read
$0.275
Cache write
free
Batch and priority prices
US dollars per million tokens · per listing
| Price | Listing | $/M |
|---|---|---|
| Batch input | openai | $0.55 |
| Batch output | openai | $2.20 |
| Priority input | openai | $2.00 |
| Priority output | openai | $8.00 |
Batch input is 50% below the interactive input price at the same listing.
Quality
12 benchmarks · 3 sources
Composite
55th
percentile of 334 scored models
Rank
152 / 334
±0.05 sd
Evidence
12 × 3
3 or more benchmarks, from 2 or more sources
Effort range
—
one configuration only
coding61st2/3 bench
reasoning56th3/4 bench
math54th6/7 bench
knowledge16th1/1 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| MATH level 5 | 97.8 ±0.3 | default | Epoch AI | 2025-04-16 |
| Aider polyglot | 72.0 | high · aider diff | Aider | 2025-04-16 |
| GPQA diamond | 77.5 ±3.0 | default · 3 runs | Epoch AI | 2026-08-07 |
| OTIS Mock AIME 2024-2025 | 70.9 ±6.7 | default · 3 runs | Epoch AI | 2026-08-07 |
| Chess Puzzles | 20.0 ±4.0 | default · 3 runs | Epoch AI | 2026-08-07 |
| FrontierMath-v1 | 22.4 ±1.8 | default · 6 runs | Epoch AI | 2025-11-16 |
| SWE-Bench verified | 54.8 | default · PatchPilot-v1.1 · 2 runs | SWE-bench | — |
| FrontierMath-Tier-4-2025-07-01 | 2.1 ±0.0 | default · 4 runs | Epoch AI | 2025-08-07 |
| FrontierMath-Tiers-1-3-v2 | 27.0 ±2.2 | default · 3 runs | Epoch AI | 2026-08-27 |
| FrontierMath-Tier-4-v2 | 4.9 ±3.4 | default | Epoch AI | 2026-06-11 |
| Mystery Game Puzzles | 5.0 ±2.2 | default | Epoch AI | 2026-08-27 |
| SimpleQA Verified | 19.2 ±1.3 | default · 2 runs | Epoch AI | 2026-08-27 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 22 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| openaivendor | $1.10 | $4.40 | $0.275 | — | 200K | 100K | 2026-10-23 |
| poe | $0.99 | $4.00 | $0.25 | — | 200K | 100K | — |
| replicate · replicate | $1.00 | $4.00 | — | — | — | — | — |
| abacus | $1.10 | $4.40 | — | — | 200K | 100K | — |
| azure | $1.10 $1.21 | $4.40 $4.84 | $0.275 $0.31 | — | 200K | 100K | 2026-11-19 |
| azure-cognitive-services | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| cloudflare-ai-gateway | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| edenai | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| helicone | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| impossibl | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| jiekou | $1.10 | $4.40 | — | — | 200K | 100K | — |
| kilo | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| llmgateway | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| llmgateway-providers | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| merge-gateway | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| nano-gpt | $1.10 | $4.40 | $0.55 | — | 200K | 100K | — |
| nearai | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| openrouter | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| vercel | $1.10 | $4.40 | $0.275 | — | 200K | 100K | — |
| vercel_ai_gateway · vercel-ai-gateway | $1.10 | $4.40 | $0.275 | free | 200K | 100K | — |
| anyapi | — | — | — | — | 200K | 100K | — |
| model-oracle-ai | — | — | — | — | 200K | 100K | — |
Price history
append-only observations · a listing writes a row only when its price moves
| Date | Host | Field | $/M |
|---|---|---|---|
| 2026-09-02 | openrouter | Cache read | $0.275 |
| 2026-09-06 | openrouter | Cache read | $0.275 |
| 2026-09-02 | openrouter | Input | $1.10 |
| 2026-09-06 | openrouter | Input | $1.10 |
| 2026-09-02 | openrouter | Output | $4.40 |
| 2026-09-06 | openrouter | Output | $4.40 |
More from OpenAI
most-hosted first