Compare · ModelsLive · 2 picked · head to head
Gemma 4 26B A4B vs o3
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
o3 wins on 14/15 benchmarks
o3 wins 14 of 15 shared benchmarks. Leads in speed · knowledge · general.
Category leads
speed·o3knowledge·o3general·o3math·o3coding·o3
Head to head · from the data · data as of 2026-11-22
- Benchmarks · Gemma 4 26B A4B leads on 1 of 15 shared benchmarks (Artificial Analysis · IFBench), o3 on 14 (Artificial Analysis · CritPt, Artificial Analysis · GPQA Diamond, Artificial Analysis · Humanity's Last Exam, Artificial Analysis · Long Context Reasoning, 10 more)
- Price · Gemma 4 26B A4B input $0.07/M vs o3 $2.00/M, Gemma 4 26B A4B is 97% cheaper (OpenRouter list price, as of 2026-11-22)
- Cheapest provider · Gemma 4 26B A4B at Darkbloom $0.05/M input (of 15 offers) as of 2026-11-22 · o3 at OpenAI $2.00/M input (1 offer) as of 2026-11-22
- Context · Gemma 4 26B A4B 262K tokens vs o3 200K tokens
- Released · Gemma 4 26B A4B Apr 3, 2026 vs o3 Apr 16, 2025 · 12 months apart
- Open weights · Gemma 4 26B A4B has open weights (on Hugging Face as google/gemma-4-26B-A4B-it)
- Price history · o3 OpenRouter input list price had no move of 5% or more in the 90 days to 2026-11-22
Sources · scores from public leaderboards, list prices and provider offers from OpenRouter, price moves from the BenchGecko changelog (5% or more, held in the next daily check), Gecko tests run by BenchGecko.
Hype vs Reality
Attention vs performance
Gemma 4 26B A4B
#166 by perf·#8 by attention
o3
#124 by perf·no signal
Best value
Gemma 4 26B A4B
30.9x better value than o3
Gemma 4 26B A4B
317.3 pts/$
$0.15/M
o3
10.3 pts/$
$5.00/M
Vendor risk
Who is behind the model
Google DeepMind
$3.26T·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
15 benchmarks · 2 models
Gemma 4 26B A4B o3
Artificial Analysis · CritPt
o3 leads by +1.1
Gemma 4 26B A4B
0.0
o3
1.1
Artificial Analysis · GPQA Diamond
o3 leads by +3.5
Gemma 4 26B A4B
79.2
o3
82.7
Artificial Analysis · Humanity's Last Exam
o3 leads by +0.8
Gemma 4 26B A4B
19.3
o3
20.1
Artificial Analysis · IFBench
Gemma 4 26B A4B leads by +1.0
Gemma 4 26B A4B
72.4
o3
71.4
Artificial Analysis · Long Context Reasoning
o3 leads by +9.0
Gemma 4 26B A4B
65.7
o3
74.7
Artificial Analysis · MMMU Pro
o3 leads by +0.9
Gemma 4 26B A4B
69.2
o3
70.1
Artificial Analysis · Quality Index
o3 leads by +3.5
Gemma 4 26B A4B
16.7
o3
20.2
Artificial Analysis · tau2-Bench Telecom
o3 leads by +37.1
Gemma 4 26B A4B
43.6
o3
80.7
Artificial Analysis · Terminal-Bench Hard
o3 leads by +23.5
Gemma 4 26B A4B
13.6
o3
37.1
Chess Puzzles
o3 leads by +32.6
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Gemma 4 26B A4B
1.1
o3
33.7
Dtbench
o3 leads by +16.5
Gemma 4 26B A4B
58.2
o3
74.7
GPQA diamond
o3 leads by +8.8
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemma 4 26B A4B
67.0
o3
75.8
Lmca
o3 leads by +11.8
Gemma 4 26B A4B
34.9
o3
46.7
OTIS Mock AIME 2024-2025
o3 leads by +2.2
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemma 4 26B A4B
82.2
o3
84.4
WeirdML
o3 leads by +17.3
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Gemma 4 26B A4B
35.2
o3
52.4
Full benchmark table
| Benchmark | Gemma 4 26B A4B | o3 |
|---|---|---|
Artificial Analysis · CritPt | 0.0 | 1.1 |
Artificial Analysis · GPQA Diamond | 79.2 | 82.7 |
Artificial Analysis · Humanity's Last Exam | 19.3 | 20.1 |
Artificial Analysis · IFBench | 72.4 | 71.4 |
Artificial Analysis · Long Context Reasoning | 65.7 | 74.7 |
Artificial Analysis · MMMU Pro | 69.2 | 70.1 |
Artificial Analysis · Quality Index | 16.7 | 20.2 |
Artificial Analysis · tau2-Bench Telecom | 43.6 | 80.7 |
Artificial Analysis · Terminal-Bench Hard | 13.6 | 37.1 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 1.1 | 33.7 |
Dtbench | 58.2 | 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. | 67.0 | 75.8 |
Lmca | 34.9 | 46.7 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 82.2 | 84.4 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 35.2 | 52.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.07 | $0.23 | 262K tokens (~131 books) | $1.07 | |
| $2.00 | $8.00 | 200K tokens (~100 books) | $35.00 |
Gemma 4 26B A4B vs o3 · questions answered from the data
Gemma 4 26B A4B leads on 1 of 15 shared benchmarks (Artificial Analysis · IFBench), o3 on 14 (Artificial Analysis · CritPt, Artificial Analysis · GPQA Diamond, Artificial Analysis · Humanity's Last Exam, Artificial Analysis · Long Context Reasoning, 10 more)