Compare · ModelsLive · 3 picked · head to head
Claude Mythos Preview vs Gemini 4 Argon vs GPT-6 Astra
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-6 Astra wins on 10/19 benchmarks
GPT-6 Astra wins 10 of 19 shared benchmarks. Leads in speed · coding · math.
Category leads
knowledge·Gemini 4 Argonspeed·GPT-6 Astraagentic·Gemini 4 Argoncoding·GPT-6 Astramath·GPT-6 Astrageneral·GPT-6 Astra
Hype vs Reality
Attention vs performance
Claude Mythos Preview
#4 by perf·no signal
Gemini 4 Argon
#11 by perf·#1 by attention
GPT-6 Astra
#10 by perf·#5 by attention
Best value
Gemini 4 Argon
5.0x better value than GPT-6 Astra
Claude Mythos Preview
n/a
no price
Gemini 4 Argon
12.5 pts/$
$6.00/M
GPT-6 Astra
2.5 pts/$
$30.00/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
Google DeepMind
$4.27T·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
19 benchmarks · 3 models
Claude Mythos PreviewGemini 4 ArgonGPT-6 Astra
GPQA diamond
Gemini 4 Argon leads by +0.7
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Mythos Preview
94.5
Gemini 4 Argon
95.2
GPT-6 Astra
94.4
Artificial Analysis · CritPt
GPT-6 Astra leads by +4.6
Gemini 4 Argon
27.1
GPT-6 Astra
31.7
Artificial Analysis · GDPval
Gemini 4 Argon leads by +2.6
Gemini 4 Argon
56.2
GPT-6 Astra
53.6
Artificial Analysis · Humanity's Last Exam
Gemini 4 Argon leads by +2.4
Gemini 4 Argon
57.1
GPT-6 Astra
54.7
Artificial Analysis · Long Context Reasoning
GPT-6 Astra leads by +1.0
Gemini 4 Argon
79.7
GPT-6 Astra
80.7
Artificial Analysis · Quality Index
GPT-6 Astra leads by +0.1
Gemini 4 Argon
52.6
GPT-6 Astra
52.7
Artificial Analysis · SciCode
Gemini 4 Argon leads by +5.3
Gemini 4 Argon
61.8
GPT-6 Astra
56.5
APEX-Agents
Gemini 4 Argon leads by +17.7
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Gemini 4 Argon
82.4
GPT-6 Astra
64.7
Chess Puzzles
GPT-6 Astra leads by +15.8
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Gemini 4 Argon
56.9
GPT-6 Astra
72.6
Frontiercode
GPT-6 Astra leads by +5.7
Gemini 4 Argon
47.5
GPT-6 Astra
53.3
FrontierMath-Tier-4-v2-Private
GPT-6 Astra leads by +19.5
Gemini 4 Argon
78.0
GPT-6 Astra
97.6
FrontierMath-Tiers-1-3-v2-Private
GPT-6 Astra leads by +7.7
Gemini 4 Argon
86.0
GPT-6 Astra
93.7
Frontierswe
GPT-6 Astra leads by +10.6
Gemini 4 Argon
55.0
GPT-6 Astra
65.5
Furniture Assembly
GPT-6 Astra leads by +10.7
Gemini 4 Argon
60.7
GPT-6 Astra
71.4
HLE
Claude Mythos Preview leads by +4.3
HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation. Claude Opus 4.7 scores 46.9% without tools and 54.7% with tools · making it one of the few benchmarks where the top score is below 60%.
Claude Mythos Preview
56.8
GPT-6 Astra
52.5
Mystery Game Puzzles
GPT-6 Astra leads by +22.0
Gemini 4 Argon
60.3
GPT-6 Astra
82.4
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemini 4 Argon
100.0
GPT-6 Astra
100.0
Proofbench
Gemini 4 Argon
99.0
GPT-6 Astra
99.0
SimpleQA Verified
Gemini 4 Argon leads by +0.4
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Gemini 4 Argon
76.0
GPT-6 Astra
75.6
Full benchmark table
| Benchmark | Claude Mythos Preview | Gemini 4 Argon | GPT-6 Astra |
|---|---|---|---|
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 94.5 | 95.2 | 94.4 |
Artificial Analysis · CritPt | — | 27.1 | 31.7 |
Artificial Analysis · GDPval | — | 56.2 | 53.6 |
Artificial Analysis · Humanity's Last Exam | — | 57.1 | 54.7 |
Artificial Analysis · Long Context Reasoning | — | 79.7 | 80.7 |
Artificial Analysis · Quality Index | — | 52.6 | 52.7 |
Artificial Analysis · SciCode | — | 61.8 | 56.5 |
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | — | 82.4 | 64.7 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | — | 56.9 | 72.6 |
Frontiercode | — | 47.5 | 53.3 |
FrontierMath-Tier-4-v2-Private | — | 78.0 | 97.6 |
FrontierMath-Tiers-1-3-v2-Private | — | 86.0 | 93.7 |
Frontierswe | — | 55.0 | 65.5 |
Furniture Assembly | — | 60.7 | 71.4 |
HLE HLE (Humanity's Last Exam) · a reasoning benchmark designed to be the hardest public evaluation of AI. Questions span mathematics, physics, philosophy, and logic · curated to be at or beyond the frontier of human expert capability. Tested with and without tool augmentation. Claude Opus 4.7 scores 46.9% without tools and 54.7% with tools · making it one of the few benchmarks where the top score is below 60%. | 56.8 | — | 52.5 |
Mystery Game Puzzles | — | 60.3 | 82.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | — | 100.0 | 100.0 |
Proofbench | — | 99.0 | 99.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | — | 76.0 | 75.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| — | — | 1.0M tokens (~500 books) | — | |
| $2.00 | $10.00 | 1.0M tokens (~524 books) | $40.00 | |
| $10.00 | $50.00 | 1.1M tokens (~525 books) | $200.00 |