Across 4 comparable benchmarks, Gemini 3.0 Flash has better reported results on 3 and worse on 1.
Moderate evidence4 of 4 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | Gemini 2.5 Deep Think | Gemini 3.0 Flash |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | $0.400 | $0.500 |
| Output / 1M | $1.60 | $3.00 |
| Blended (4:1) | $0.640 | $0.676 |
| Status | Available | Preview |
| Released | August 1, 2025 | December 17, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 2 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Gemini 3.0 Flash ahead on 1 of 1 counted
No fixed-scale average — see the rows in Benchmark results
Split 1–1 of 2 counted
No shared comparable results for Abstract reasoning, Science & health, Computer & tool use, Research & search, Long context, Multimodal, Professional work & productivity, Cybersecurity.
Published scores, with matching modes where available. Gaps show Gemini 2.5 Deep Think relative to Gemini 3.0 Flash. Reading the results
| Benchmark | Gemini 2.5 Deep Think | Gemini 3.0 Flash | Gap |
|---|---|---|---|
| LiveCodeBenchAlgorithmic Coding · Pass @K | 87.6Deep Thinking · no tools | 90.8≠Thinking · no tools | 0.96×#16 vs #7 of 113 |
| AIME2025Mathematics · Accuracy | 99.2Deep Thinking · no tools | 99.7≠Thinking · tools | −0.5 pts#11 vs #8 of 86 |
| FrontierMath - Tier 4Mathematics · Accuracy | 10.4Normal · no tools | 4.2Normal · no tools | +6.2 pts#15 vs #21 of 50 |
| HLEKnowledge Exams · Accuracy | 34.8Deep Thinking · no tools | 43.5≠Thinking · tools | −8.7 pts#83 vs #71 of 151 |
Algorithmic Coding · Pass @K
Mathematics · Accuracy
Mathematics · Accuracy
Knowledge Exams · Accuracy
A dash means no result is available.
| Benchmark | Gemini 2.5 Deep Think | Gemini 3.0 Flash |
|---|---|---|
| AIME 2026 | — | 96.7 |
| CritPt | — | 8.6 |
| FrontierMath | 29.0 | — |
| FrontierMath Tier 4 v2 | — | 17.1 |
| FrontierMath v2 | — | 51.2 |
| GPQA Diamond | — | 90.4 |
| IMO-ProofBench | 37.6 | — |
| MMLU-Pro | — | 89.0 |
| Code Migration | — | 6.4 |
| GSO | — | 9.8 |
| SWE-bench Multilingual | — | 72.7 |
| SWE-Bench Pro - Public | — | 49.6 |
| SWE-bench Verified | — | 75.8 |
| Terminal Bench 2.0 | — | 47.6 |
| Terminal Bench Hard | — | 38.6 |
| Terminal-Bench 2.1 | — | 58.0 |
| Vibe Code Bench v1.1 | — | 20.2 |
| WeirdML v2 | — | 61.6 |
| Claw Bench | — | 85.7 |
| MCP-Atlas | — | 62.0 |
| Pinch Bench | — | 85.2 |
| PinchBench v2 | — | 72.1 |
| ECI | — | 152 |
| LiveBench | — | 72.4 |
| Text Arena (Coding) | — | 1454 |
| IF Bench | — | 78.0 |
| GeoBench ACW | — | 88.0 |
| MMMU-Pro | — | 79.9 |
| VPCT | — | 72.6 |
| EMB (Excel Modeling) | — | 26.0 |
| Finance Agent v2 | — | 42.5 |
| Harvey's Legal Agent Benchmark | — | 0.0 |
| Legal Research Bench | — | 18.3 |
| SAGE | — | 51.9 |
| Tax Agent Bench | — | 11.8 |
| τ²-Bench | — | 90.2 |
| τ²-Bench - Telecom | — | 91.2 |
| τ³-Banking | — | 27.3 |
| ARC-AGI-1 | — | 84.7 |
| ARC-AGI-2 | — | 33.6 |
| BALROG | — | 48.1 |
| SimpleBench | — | 61.1 |
| SimpleQA | — | 68.7 |
| MedCode | — | 55.9 |
| MedScribe | — | 69.9 |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
Gemini 2.5 Deep Think costs5.3%lessthan Gemini 3.0 Flash per million tokens.Using standard rates at this token mix.
Gemini 3.0 Flash: Standard pricing. Google Gemini API official pricing (ai.google.dev/gemini-api/docs/pricing) verified 2026-11-13; paid tier; standard tier; text input Google Gemini API official pricing (ai.google.dev/gemini-api/docs/pricing) verified 2026-11-13; paid tier; standard tier; output
| Price / 1M tokens | Gemini 2.5 Deep Think | Gemini 3.0 Flash | Ratio |
|---|---|---|---|
| Input | $0.40 | $0.50 | 0.80× |
| Output | $1.60 | $3.00 | 0.53× |
| Cached input | — | $0.05 | — |
| Cache write | — | — | — |
| Blended | $0.64 | $0.676 | 0.95× |
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 2.5 Deep Think | Gemini 3.0 Flash |
|---|---|---|
| AIME 2026 | Not reported | |
| ARC-AGI-1 | Not reported | |
| ARC-AGI-2 | Not reported | |
| Code Migration | Not reported | |
| EMB (Excel Modeling) | Not reported | |
| Finance Agent v2 | Not reported | |
| Legal Research Bench | Not reported | |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| SAGE | Not reported | |
| Tax Agent Bench | Not reported | $0.1007 / task Average · Accuracy (%, rubric × citation quality): 11.8 Thinking High (With Tools) |
| Vibe Code Bench v1.1 | Not reported |
| Attribute | Gemini 2.5 Deep Think | Gemini 3.0 Flash |
|---|---|---|
| Availability | Available | Preview |
| Provider | Google Deep Mind | Google Deep Mind |
| Model type | Chat model | Chat model |
| First release | August 1, 2025 | December 17, 2025 |
| Context window | 1M tokens | 1M tokens |
| Max output | 16K tokens | 64K tokens |
| Architecture | Undisclosed | Undisclosed |
| Documented capabilities | Reasoning, Multilingual | Reasoning, Multilingual |
| Input modalities | Text | Text, Image, Audio, Video |
| Output modalities | Text | 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 2.5 Deep Think or Gemini 3.0 Flash? 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.