Compare · ModelsLive · 2 picked · head to head
gpt-oss-20b vs o3
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
o3 wins on 18/19 benchmarks
o3 wins 18 of 19 shared benchmarks. Leads in speed · knowledge · general.
Category leads
speed·o3knowledge·o3general·o3language·o3math·o3reasoning·o3coding·o3
Head to head · from the data · data as of 2026-11-22
- Benchmarks · gpt-oss-20b leads on 1 of 19 shared benchmarks (Artificial Analysis · CritPt), o3 on 18 (Artificial Analysis · GPQA Diamond, Artificial Analysis · Humanity's Last Exam, Artificial Analysis · IFBench, Artificial Analysis · Long Context Reasoning, 14 more)
- Price · gpt-oss-20b input $0.02/M vs o3 $2.00/M, gpt-oss-20b is 99% cheaper (OpenRouter list price, as of 2026-11-22)
- Cheapest provider · gpt-oss-20b at Darkbloom $0.02/M input (of 12 offers) as of 2026-11-22 · o3 at OpenAI $2.00/M input (1 offer) as of 2026-11-22
- Context · gpt-oss-20b 131K tokens vs o3 200K tokens
- Released · gpt-oss-20b Aug 5, 2025 vs o3 Apr 16, 2025 · 4 months apart
- Open weights · gpt-oss-20b has open weights (on Hugging Face as openai/gpt-oss-20b)
- Price history · o3 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-22
Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.
Hype vs Reality
Attention vs performance
gpt-oss-20b
#107 by perf·no signal
o3
#124 by perf·no signal
Best value
gpt-oss-20b
101.7x better value than o3
gpt-oss-20b
1043.1 pts/$
$0.05/M
o3
10.3 pts/$
$5.00/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
19 benchmarks · 2 models
gpt-oss-20bo3
Artificial Analysis · CritPt
gpt-oss-20b leads by +0.3
gpt-oss-20b
1.4
o3
1.1
Artificial Analysis · GPQA Diamond
o3 leads by +13.9
gpt-oss-20b
68.8
o3
82.7
Artificial Analysis · Humanity's Last Exam
o3 leads by +9.1
gpt-oss-20b
11.0
o3
20.1
Artificial Analysis · IFBench
o3 leads by +6.3
gpt-oss-20b
65.1
o3
71.4
Artificial Analysis · Long Context Reasoning
o3 leads by +40.0
gpt-oss-20b
34.7
o3
74.7
Artificial Analysis · Quality Index
o3 leads by +11.2
gpt-oss-20b
9.0
o3
20.2
Artificial Analysis · tau2-Bench Telecom
o3 leads by +20.5
gpt-oss-20b
60.2
o3
80.7
Artificial Analysis · Terminal-Bench Hard
o3 leads by +26.5
gpt-oss-20b
10.6
o3
37.1
Chess Puzzles
o3 leads by +33.7
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
gpt-oss-20b
0.0
o3
33.7
Dtbench
o3 leads by +28.0
gpt-oss-20b
46.7
o3
74.7
GPQA diamond
o3 leads by +26.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
gpt-oss-20b
49.7
o3
75.8
HELM · GPQA
o3 leads by +15.9
gpt-oss-20b
59.4
o3
75.3
HELM · IFEval
o3 leads by +13.7
gpt-oss-20b
73.2
o3
86.9
HELM · MMLU-Pro
o3 leads by +11.9
gpt-oss-20b
74.0
o3
85.9
HELM · Omni-MATH
o3 leads by +14.9
gpt-oss-20b
56.5
o3
71.4
HELM · WildBench
o3 leads by +12.4
gpt-oss-20b
73.7
o3
86.1
Lmca
o3 leads by +29.6
gpt-oss-20b
17.1
o3
46.7
OTIS Mock AIME 2024-2025
o3 leads by +19.2
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
gpt-oss-20b
65.2
o3
84.4
WeirdML
o3 leads by +11.5
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
gpt-oss-20b
40.9
o3
52.4
Full benchmark table
| Benchmark | gpt-oss-20b | o3 |
|---|---|---|
Artificial Analysis · CritPt | 1.4 | 1.1 |
Artificial Analysis · GPQA Diamond | 68.8 | 82.7 |
Artificial Analysis · Humanity's Last Exam | 11.0 | 20.1 |
Artificial Analysis · IFBench | 65.1 | 71.4 |
Artificial Analysis · Long Context Reasoning | 34.7 | 74.7 |
Artificial Analysis · Quality Index | 9.0 | 20.2 |
Artificial Analysis · tau2-Bench Telecom | 60.2 | 80.7 |
Artificial Analysis · Terminal-Bench Hard | 10.6 | 37.1 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 0.0 | 33.7 |
Dtbench | 46.7 | 74.7 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 49.7 | 75.8 |
HELM · GPQA | 59.4 | 75.3 |
HELM · IFEval | 73.2 | 86.9 |
HELM · MMLU-Pro | 74.0 | 85.9 |
HELM · Omni-MATH | 56.5 | 71.4 |
HELM · WildBench | 73.7 | 86.1 |
Lmca | 17.1 | 46.7 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 65.2 | 84.4 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 40.9 | 52.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.02 | $0.09 | 131K tokens (~66 books) | $0.34 | |
| $2.00 | $8.00 | 200K tokens (~100 books) | $35.00 |
gpt-oss-20b vs o3 · questions answered from the data
gpt-oss-20b leads on 1 of 19 shared benchmarks (Artificial Analysis · CritPt), o3 on 18 (Artificial Analysis · GPQA Diamond, Artificial Analysis · Humanity's Last Exam, Artificial Analysis · IFBench, Artificial Analysis · Long Context Reasoning, 14 more)