Across 10 comparable benchmarks, Gemini 4 Argon has better reported results on 10 and worse on 0. 3 further shared results are listed below, each with the reason it is not counted.
Strong evidence10 of 13 shared results countedHow we compare
Gemini 3.0 Pro (Preview 11-2025) is no longer available through its official API. API prices for this version are not shown.
Price basis: Current effective rates. The price section below shows the applicable periods.
| Fact | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | β | $2.00 |
| Output / 1M | β | $10.00 |
| Blended (4:1) | β | $1.12 |
| Status | Retired | Available |
| Released | November 18, 2025 | October 13, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates β set your own mix.
Mean of 7 shared benchmarks grouped by capability on a 0β100 scale. How averages are built
Gemini 4 Argon ahead on 2 of 2 counted
Gemini 4 Argon ahead on 2 of 2 counted
Gemini 4 Argon ahead on 4 of 4 counted
Gemini 4 Argon ahead on 2 of 2 counted
No shared comparable results for Mathematics, Computer & tool use, Research & search, Long context, Multimodal, Cybersecurity.
Published scores, with matching modes where available. Gaps show Gemini 3.0 Pro (Preview 11-2025) relative to Gemini 4 Argon. Reading the results
| Benchmark | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon | Gap |
|---|---|---|---|
| APEX-SWERepository Engineering Β· Pass@1 | 26.8Thinking High Β· tools | 49.2Thinking High Β· tools | β22.4 pts#54 vs #15 of 61 |
| Vibe Code Bench v1.1Code Generation & Editing Β· Pass rate (%) | 14.3Thinking High Β· tools | 91.9Thinking High Β· tools | β77.6 pts#96 vs #4 of 121 |
| ARC-AGI-1Abstract Generalization Β· Score (%) | 87.5Parallel Thinking Β· no tools | 98.5β Thinking Medium Β· no tools | β11.0 pts#40 vs #1 of 87 |
| ARC-AGI-2Abstract Generalization Β· Score (% solved); cost per task (USD)Not counted Β· Speed and cost telemetry is reported, not scored | 45.1Parallel Thinking Β· no tools | 91.3β Thinking High Β· no tools | 0.49Γ#46 vs #8 of 80 |
| SimpleBenchCommonsense Β· Score (AVG@5) | 76.4Thinking Β· no tools | 91.7β Reported best | 0.83Γ#19 vs #4 of 106 |
| Vending-Bench 2Logical Planning Β· Money Balance (USD)Not counted Β· Speed and cost telemetry is reported, not scored | 5478Unspecified Reasoning Β· tools | 13718Unspecified Reasoning Β· tools | 0.40Γ#33 vs #6 of 64 |
| CritPtScientific Reasoning Β· Score | 9.1Thinking High Β· no tools | 27.1Thinking High Β· no tools | 0.34Γ#58 vs #17 of 134 |
| GPQA DiamondScientific Reasoning Β· Accuracy | 93.8Parallel Thinking Β· no tools | 96.4β Thinking High Β· no tools | β2.6 pts#23 vs #2 of 200 |
| MedCodeClinical Workflows Β· Accuracy (%) | 52.2Thinking High Β· tools | 58.8Thinking High Β· tools | β6.6 pts#14 vs #3 of 97 |
| MedScribeClinical Workflows Β· Rubric pass rate (%) | 72.0Thinking High Β· tools | 87.4Thinking High Β· tools | β15.4 pts#84 vs #17 of 99 |
| Creative WritingWriting | 1524Normal Β· no tools | 1758Normal Β· no tools | 0.87Γ#55 vs #28 of 121 |
| SAGEService Workflows Β· Accuracy (%) | 47.6Thinking High Β· tools | 53.6Thinking High Β· tools | β6.0 pts#31 vs #5 of 82 |
| HLEKnowledge Exams Β· AccuracyNot counted Β· Different or unrecorded test environments β excluded from the tally | 45.8Thinking High Β· toolsΒ· legacy-default | 57.1β Thinking High Β· no tools Β· text onlyΒ· text-only | β#63 vs #25 of 151 |
Repository Engineering Β· Pass@1
Code Generation & Editing Β· Pass rate (%)
Abstract Generalization Β· Score (%)
Abstract Generalization Β· Score (% solved); cost per task (USD)
Not counted Β· Speed and cost telemetry is reported, not scored
Commonsense Β· Score (AVG@5)
Logical Planning Β· Money Balance (USD)
Not counted Β· Speed and cost telemetry is reported, not scored
Scientific Reasoning Β· Score
Scientific Reasoning Β· Accuracy
Clinical Workflows Β· Accuracy (%)
Clinical Workflows Β· Rubric pass rate (%)
Writing
Service Workflows Β· Accuracy (%)
Knowledge Exams Β· Accuracy
Not counted Β· Different or unrecorded test environments β excluded from the tally
A dash means no result is available.
Both models publish peak and off-peak prices. Each period is compared separately using the same token mix; hours follow each providerβs schedule.
Drag the handle above the bar for cache hits, the one below it for the input : output mix Β· 5% steps
Blended price = effective input rate at 90% cache hits Γ 80% + output rate Γ 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
Gemini 3.0 Pro (Preview 11-2025): Listed text rates; pricing tier not confirmed.
Gemini 4 Argon: Listed text rates; pricing tier not confirmed.
| Price / 1M tokens | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon | Ratio |
|---|---|---|---|
| Input | β | $2.00 | β |
| Output | β | $10.00 | β |
| Cached input | β | $0.10 | β |
| Cache write | β | β | β |
| Blended | β | $2.23 | β |
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
Gemini 3.0 Pro (Preview 11-2025): Listed text rates; pricing tier not confirmed.
Gemini 4 Argon: Standard pricing. Google Gemini API official pricing (ai.google.dev/gemini-api/docs/pricing) verified 2026-10-21; paid tier; standard tier; scheduled price starting 2026-12-01
| Price / 1M tokens | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon | Ratio |
|---|---|---|---|
| Input | β | $4.00 | β |
| Output | β | $20.00 | β |
| Cached input | β | $0.20 | β |
| Cache write | β | β | β |
| Blended | β | $4.46 | β |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
| Benchmark | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon |
|---|---|---|
| AA Coding Agent Index v1.5 | Not reported | |
| AA Coding Agent Index v2.0 | Not reported | |
| AA Intelligence Index v4.3 | Not reported | |
| AA Intelligence Index v5.0.1 | Not reported | |
| AA-Omniscience | Not reported | |
| Agent Arena | Not reported | $0.694 / task Median Β· Net improvement index (percentage points vs. average model): 10.7 Thinking High (With Tools + Internet) |
| Agent Arena Index | Not reported | $0.6569 / task Median Β· Agent Arena Index rating (index points): 1229 Thinking High (With Tools + Internet) |
| AIME 2026 | Not reported | |
| ARC-AGI-1 | Not reported | |
| ARC-AGI-2 | Not reported | |
| AutomationBench | Not reported | $1.70 / task Average Β· Pass Rate: 51.3 Thinking High (With Tools) Β· zapier-private-heldout $1.54 / task Average Β· Pass Rate: 50.1 Thinking Medium (With Tools) Β· zapier-private-heldout |
| AutomationBench-AA | Not reported | $0.6087 / task Average Β· Guardrail-safe objective completion rate: 77.5 Thinking High (With Tools) |
| CritPt | Not reported | |
| CursorBench 4.0 | Not reported | $4.57 / task Average Β· Task score: 52.9 Thinking High (With Tools) Β· cursorbench-4-0-cursor-agent $4.09 / task Average Β· Task score: 50.6 Thinking Low (With Tools) Β· cursorbench-4-0-cursor-agent $4.39 / task Average Β· Task score: 52.8 Thinking Medium (With Tools) Β· cursorbench-4-0-cursor-agent |
| CWE-bench v1 | Not reported | |
| FrontierCode 2.0 | Not reported | $6.514 / task Average Β· Weighted rubric score: 26.3 Thinking High (With Tools + Internet) $3.977 / task Average Β· Weighted rubric score: 21.5 Thinking Low (With Tools + Internet) $5.688 / task Average Β· Weighted rubric score: 23.4 Thinking Medium (With Tools + Internet) |
| FrontierCS 2 | Not reported | |
| FrontierSWE v2 | Not reported | |
| GDPx | Not reported | $12.79 / task Average Β· GDPx β Score: 7.3 Thinking High (With Tools) Β· overall/score |
| HLE | Not reported | $0.1764 / task Average Β· Accuracy: 57.1 Thinking High (No Tools) Β· Text only Β· text-only |
| LiveBench 2.0 | Not reported | $1.5159 / task Average Β· Overall score (equal-weight mean of 7 suite scores, 0-100): 84.1 Thinking High (With Tools) |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| MysteryMechanism | Not reported | |
| ProofBench v1.1 (Lean 4) | Not reported | |
| Public Benefits Bench v1.1 | Not reported | |
| SAGE | Not reported | |
| SciCode | Not reported | |
| Tax Agent Bench | Not reported | $3.4736 / task Average Β· Accuracy (%, rubric Γ citation quality): 44.6 Thinking High (With Tools) |
| Terminal-Bench 4.0 | Not reported | $16.8269 / task Average Β· Resolution rate (%): 55.6 Thinking High (With Tools) Β· vals-effort-study $13.4495 / task Average Β· Resolution rate (%): 52.5 Thinking Low (With Tools) Β· vals-effort-study 2 more settings$13.3909 / task Average Β· Resolution rate (%): 56.6 Thinking Medium (With Tools) Β· vals-effort-study $17.64 / task Average Β· Resolution rate (%): 57.6 Unspecified Reasoning (With Tools) Β· vals-mini-swe |
| Terminal-Bench 4.1 | Not reported | |
| Terminal-Bench-Science 0.1 | Not reported | |
| Vals Index v2.1 | Not reported | |
| Vals Index v2.2 | Not reported | |
| Vibe Code Bench 1-100 | Not reported | |
| Vibe Code Bench v1.1 | Not reported | |
| WeirdML v3 | Not reported | $16.2075 / task Average Β· Official score (0-1; higher is better): 0.2 Thinking High (With Tools) |
| Attribute | Gemini 3.0 Pro (Preview 11-2025) | Gemini 4 Argon |
|---|---|---|
| Availability | Retired | Available |
| Provider | Google Deep Mind | Google Deep Mind |
| Model type | Multimodal model | Multimodal model |
| First release | November 18, 2025 | October 13, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | 64K tokens | 1M tokens |
| Architecture | Undisclosed | Undisclosed |
| Documented capabilities | Reasoning, Multilingual | Reasoning, Coding, Tool use, Computer use, Long context |
| Input modalities | Text, Image, Audio, Video | Text, Image, Audio, Video |
| Output modalities | Text, Audio | Text |
| Weights | Closed / API only | Closed / API only |
| Commercial use | Proprietary | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Gemini 3.0 Pro (Preview 11-2025) or Gemini 4 Argon? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.