gemini-3.5-flash-high: benchmark scores, price and ranking
gemini-3.5-flash-high from Google ranks #74 of 743 LLMs on the LLMs Tiger aggregate benchmark score (52.2), based on 11 benchmark results captured up to 2026-10-13.
- Organisation
- Type
- Fast / low-cost (flash) model
- Open weights
- No
- Released
- 2026-05-19 (source: epoch)
- Cheapest API price
- $1.50 in / $9.00 out per million tokens (via deepinfra, observed 2026-10-13T19:47Z)
- Aggregate rank
- #74 of 743 (score 52.2)
Relative to other models, gemini-3.5-flash-high is strongest on LiveBench IF (#2 of 17) and weakest on LiveBench Data Analysis (#16 of 17).
Across the 11 benchmarks where at least ten models are scored, gemini-3.5-flash-high is in the top 10% on 0 and in the top half on 8.
Its cheapest listed input price, $1.50 per million tokens, is lower than that of 25% of the 427 models with a listed price.
gemini-3.5-flash-high benchmark results
| Benchmark | Score | Rank | Source | Captured |
|---|---|---|---|---|
| LiveBench IF | 75.6% | #2 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| LiveBench Language | 84.6% | #3 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| ARC-AGI-2 | 72.1% | #18 of 87 | ARC Prize official leaderboard | 2026-10-13 |
| LMArena Document Elo | 1,461 | #4 of 18 | LMArena | 2026-10-13 |
| LiveBench Agentic Coding | 49.0% | #6 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| LiveBench Coding | 78.2% | #6 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| LMArena Vision Elo | 1,284 | #9 of 22 | LMArena | 2026-10-05 |
| LiveBench Reasoning | 82.0% | #8 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| LiveBench Mathematics | 88.2% | #9 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
| LMArena Text Elo | 1,477 | #15 of 28 | LMArena | 2026-10-06 |
| LiveBench Data Analysis | 64.9% | #16 of 17 | LiveBench 2026_06_25 official CSV | 2026-10-13 |
About gemini-3.5-flash-high
gemini-3.5-flash-high is a smaller, lower-cost model from Google. It was released on 2026-05-19, according to Epoch AI. The cheapest API listing we track charges $1.50 per million input tokens and $9.00 per million output tokens, through deepinfra. We hold 11 benchmark results for it across 3 areas, and none is in the top tenth of its board.
Where gemini-3.5-flash-high stands
On our aggregate score, gemini-3.5-flash-high is #74 of 743 ranked models, in the top quarter. Its score is 52.2; the scale is centred on 50, so it sits 2.2 points above the typical model. That beats 90% of the ranked field. Among the 29 ranked Google models it is #7. The next one up is Gemini 3 Deep Think (2/26) at 52.4. The next one down is gemini-1.5-pro-exp-0827 at 50.9.
gemini-3.5-flash-high, benchmark by benchmark
LiveBench IF puts gemini-3.5-flash-high at 75.6%, #2 of 17. That beats 88% of the models tested, but 3.5 points short of gemini-3.1-pro-preview-high (79.1%). The results either side are gemini-3.1-pro-preview-high (79.1%) and Qwen3.7 Max (74.0%). It covers LiveBench release 2026_06_25. The board median is 65.2%, so gemini-3.5-flash-high is 10.4 points above it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
On LiveBench Language, gemini-3.5-flash-high reaches 84.6%. It ranks #3 of 17, ahead of 82% of the field, and 2.8 points behind GPT-5.5 (xhigh) at 87.4%. The results either side are gemini-3.1-pro-preview-high (85.4%) and GPT-5.4 (xhigh) (82.6%). It covers LiveBench release 2026_06_25. The board median is 77.9%, so gemini-3.5-flash-high is 6.7 points above it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
gemini-3.5-flash-high manages 72.1% on ARC-AGI-2, good for #18 of 87. GPT-6 Astra (max) leads at 95.0%, 22.9 points further up, and 79% of models sit below. The results either side are deepseek-v4.1-flash-max (72.9%) and Claude Opus 4.6 (120K, High) (69.2%). It covers abstract reasoning, ARC Prize public eval set. The board median is 22.8%, so gemini-3.5-flash-high is 49.3 points above it. Source: ARC Prize official leaderboard, captured 2026-10-13.
LMArena Document Elo puts gemini-3.5-flash-high at 1,461, #4 of 18. That beats 78% of the models tested, but 52 points short of Claude Fable 5.1 (max) (1,513). The results either side are GPT-6 Astra (max) (1,468) and Gemini 3.6 Flash (High) (1,456). It covers crowd-voted questions about documents. The board median is 1,440, so gemini-3.5-flash-high is 22 points above it. Source: LMArena, captured 2026-10-13.
On LiveBench Agentic Coding, gemini-3.5-flash-high reaches 49.0%. It ranks #6 of 17, ahead of 65% of the field, and 7.1 points behind claude-opus-4-8-xhigh-effort at 56.1%. The results either side are GPT-5.2 Codex (49.4%) and kimi-k2.6-thinking (46.9%). It covers LiveBench release 2026_06_25. The board median is 45.8%, so gemini-3.5-flash-high is 3.2 points above it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
gemini-3.5-flash-high manages 78.2% on LiveBench Coding, good for #6 of 17. GPT-5.2 Codex leads at 83.6%, 5.4 points further up, and 65% of models sit below. The results either side are kimi-k2.6-thinking (78.6%) and qwen3.6-plus (78.2%). It covers LiveBench release 2026_06_25. The board median is 76.5%, so gemini-3.5-flash-high is 1.7 points above it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
LiveBench Reasoning puts gemini-3.5-flash-high at 82.0%, #8 of 17. That beats 53% of the models tested, but 7.7 points short of claude-opus-4-8-xhigh-effort (89.7%). The results either side are Kimi K2.7 Code (82.8%) and GPT-5.4 nano (xhigh) (81.1%). It covers LiveBench release 2026_06_25. The board median is 81.1%, so gemini-3.5-flash-high is 0.9 points above it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
gemini-3.5-flash-high lands below the middle on LiveBench Mathematics: 88.2%, #9 of 17, with 47% of the field scoring higher. GPT-5.5 (xhigh) leads at 95.9%; the closest result above is GPT-5.2 Codex at 88.8%. It covers LiveBench release 2026_06_25. The median score on this board is also 88.2%. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
gemini-3.5-flash-high lands below the middle on LMArena Text Elo: 1,477, #15 of 28, with 50% of the field scoring higher. gemini-4-argon-high leads at 1,525; the closest result above is GLM-5.3 (max) at 1,478. It covers head-to-head chat preference (crowd votes). The board median is 1,478, so gemini-3.5-flash-high is 0 points below it. Source: LMArena, captured 2026-10-06.
LiveBench Data Analysis is a weak spot. gemini-3.5-flash-high scores 64.9%, #16 of 17, behind 88% of the models measured. The leader, GPT-5.5 (xhigh), reaches 81.6%. It covers LiveBench release 2026_06_25. The board median is 71.8%, so gemini-3.5-flash-high is 6.9 points below it. Source: LiveBench 2026_06_25 official CSV, captured 2026-10-13.
The full table above also lists the other 1 result, including boards with fewer than ten models, where a rank says little.
Strengths and weaknesses
Software engineering: solid, ahead of 65% of the field on average across 2 benchmarks. Its best board here is LiveBench Agentic Coding (#6 of 17) and its weakest is LiveBench Agentic Coding (#6 of 17). Reasoning, maths and science: solid, ahead of 60% of the field on average across 3 benchmarks. Its best board here is ARC-AGI-2 (#18 of 87) and its weakest is LiveBench Mathematics (#9 of 17). General knowledge and chat: solid, ahead of 60% of the field on average across 6 benchmarks. Its best board here is LiveBench IF (#2 of 17) and its weakest is LiveBench Data Analysis (#16 of 17).
Price and value
The median input price across the 427 priced models is $0.500 per million tokens, so gemini-3.5-flash-high costs more than the typical listing. Models that score within 1.5 points or better and cost less: GLM-5.2 (high) ($0.060, 51.6), DeepSeek V4 Flash 0731 (max) ($0.070, 53.1), Qwen3-Coder-30B-A3B-Instruct ($0.070, 52.7). Similar scores cost more on Qwen3.8 Max (xhigh) ($1.65) and Qwen3.8 Max (0902) (xhigh) ($1.65).
Release history
The Google model released before it was gemini-3.1-flash-lite (high) on 2026-03-03, which scores 47.7. The next one was Gemini 3.5 Flash-Lite (High) on 2026-07-21, at 48.0.
Reading these numbers
The 11 scores on this page come from 3 sources: LiveBench 2026_06_25 official CSV (7), LMArena (3), ARC Prize official leaderboard (1). 8 of them come from an official leaderboard and 3 from datasets or third-party runs. The leaderboard also lists 3 other reasoning-effort settings of this model, which redirect here. This page shows the highest.
gemini-3.5-flash-high is also listed at 3 other reasoning-effort settings; this page shows the highest-effort row, which is the one the leaderboard keeps by default.
Models ranked nearby
- #71 Qwen3.7 Max (52.3)
- #72 Intern-S2-Preview (52.3)
- #73 Darwin-397B-ZTC (52.2)
- #75 deepseek-v4.1-flash-max (52.2)
- #76 GPT-5.2 Codex (52.2)
- #77 Ornith-1.5-397B (52.2)
More from Google
- gemini-4-argon-high (#10, 59.5)
- gemini-3.1-pro-preview-high (#16, 55.3)
- gemini-2.5-pro-preview-06-05 (#35, 53.5)
- Gemini 2.5 Pro Preview 05-06 (#48, 53.0)
- Gemini 2.5 Pro Preview 03-25 (#59, 52.6)
- Gemini 3 Deep Think (2/26) (#68, 52.4)
Compare it on the live leaderboard → · How the aggregate score works