# gemini-3.8-flash decision evidence

Gemini 3.8 Flash is Google's closed-weight Flash model, generally available since 2026-09-02, with a 1,048,576-token input limit, 65,536 output tokens and text, image, video, audio and PDF input; Google notes it may use more tokens to maximize performance, especially at higher effort. In RemakeBench's four-model Flash comparison it made the fastest single-turn Rubik's Cube pass at several times the others' cost, it placed second to GPT-6 Astra on Cortex AI's robot tasks, and coding leaderboards rate its reasoning above its code (40th of 47 on AI Coding Daily's, 49/60 on kzhu's).

Verified: 2026-09-24; snapshot: 2026-09-25.

## Published model card

As the vendor publishes it for the model and its weights; these are not the limits of a hosted endpoint.

- Context window: 1,048,576 tokens
- Maximum output: 65,536 tokens
- Modalities: Text, image, video, audio and PDF input; text output
- Parameter count: Not disclosed

## Open weights and deployment

Status: Unavailable. Google publishes no Gemini 3.8 Flash weights. Its model card lists distribution through the Gemini app, Gemini Enterprise Agent Platform, Google AI Studio, the Gemini API, AI Mode and Google Antigravity.

Open-weight alternatives: [glm-5.3-flash](https://dash.chinaapi.ai/models/glm-5.3-flash/deploy), [mimo-v2.6-flash](https://dash.chinaapi.ai/models/mimo-v2.6-flash/deploy)

## Real-task reports

### [RemakeBench Launch 005: four Flash models on 14 fixtures](https://www.remakebench.com/qwen-3-8-flash-vs-deepseek-v4-1-flash-vs-glm-5-3-flash-vs-gemini-3-8-flash)

By [RemakeBench](https://www.youtube.com/@remakebench) (@remakebench). Benchmark channel that publishes per-fixture clips, costs, wall-clock times and validation notes

Published: 2026-09-18; retrieved: 2026-09-24; verified: 2026-09-24.

- Task: Fourteen fixtures spanning game building, 3D scenes, CAD, drawing and robot control, run on Qwen 3.8 Flash, DeepSeek V4.1 Flash, GLM 5.3 Flash and Gemini 3.8 Flash
- Method: Each model's attempt per fixture is shown as a video clip with its displayed cost, wall-clock time and reasoning setting; pass/fail labels are the lab's video judgments.
- Finding: No model won overall. Gemini 3.8 Flash made the fastest single-turn Rubik's Cube pass (15 min 35 s, $2.09) and was the editorial winner of the robot Eiffel Tower drawing (8 min 25 s, $1.36), but did not complete the three-move extension ($5.65) and did not deliver the required Turbofan CAD assembly.
- Limitation: Pass/fail labels are video judgments without a machine-readable validator, and single attempts do not establish reliability; the lab publishes no aggregate score or overall winner. Gemini's displayed costs are several times the other models' on most fixtures (for example $2.09 against $0.0957–$0.3414 on the single-turn cube), and its CAD presentation clip is operator work after the run.
- Single-turn cube: Pass · $2.09 · 15m 35s
- Eiffel Tower drawing: Editorial winner · $1.36 · 8m 25s

### [Cortex AI: GPT-6 Astra, Gemini 3.8 Flash and Fable 5.1 on two real robot tasks](https://cortexrobot.ai/research/llm-real-world-robotics-evals/)

By [Cortex AI Research](https://cortexrobot.ai/) (@cortexairobot). Robotics team that builds real-world robot datasets and runs the MolmoAct2 real-world evaluations

Published: 2026-09-17; retrieved: 2026-09-24; verified: 2026-09-24.

- Task: Load a sandwich into a microwave and close the door; attach a 5 mm tip to a pipette
- Method: Each model controlled a bimanual YAM robot through an overhead and two wrist cameras with depth images, at medium reasoning effort with up to 10 minutes per task; an evaluator scored each attempt with partial credit for completed steps.
- Finding: Gemini 3.8 Flash scored 13.3% on the microwave task and 30% on the pipette task, second to GPT-6 Astra (80% and 70%) and ahead of Fable 5.1 (6.7% and 20%).
- Limitation: The post does not say how many attempts each model had; scores are partial-credit judgements by the lab's evaluator, and the authors attribute the gap mainly to weaker depth perception in the lower-scoring models.
- Microwave: 13.3%
- Pipette: 30%

### [AI Coding Daily LLM Coding Leaderboard](https://aicodingdaily.com/leaderboard)

By [AI Coding Daily](https://www.youtube.com/@AICodingDaily) (@AICodingDaily). Coding-model testing channel and website that runs its own leaderboard and sells premium tutorials and channel advertising

Published: 2026-09-24; retrieved: 2026-09-24; verified: 2026-09-24.

- Task: Implement and harden production-shaped projects in several stacks, including a Laravel app, a Go service and a React/TypeScript app, from the same prompts
- Method: Each model and reasoning setting is a separate row run in a coding harness, with the average cost and time per run. Code quality is graded by a judge model against a published 100-point rubric and scaled to 20 points per project; behavioural reliability runs each of four projects five times and scores the checks every attempt fails.
- Finding: Gemini 3.8 Flash at high reasoning scored 41.67 in Antigravity, 40th of 47 rows; it was tested on a subscription, so the board shows no cost.
- Limitation: One author's benchmark whose code-quality half is graded by a judge model; rows run in different harnesses, the page is updated as models are added (read on 2026-09-24, 47 rows), and the author sells premium tutorials and advertising.
- Points (high): 41.67
- Rank: 40 of 47

### [kzhu's six-question LLM test leaderboard](https://www.techgogogo.com/llm-benchmark/)

By [AI产品狙击手 kzhu](https://www.youtube.com/@kevinzhu9305) (@kevinzhu9305). Chinese YouTube reviewer who scores models on a six-question test and publishes every score with per-question notes

Published: 2026-09-24; retrieved: 2026-09-24; verified: 2026-09-24.

- Task: Six questions scored out of 10: constrained sentence writing, two reasoning puzzles (replaced by pelican and compound-bow SVG animations in version 2.2), a browser OS with a space shooter, a 3D racing game and a Trello-style board
- Method: Each row names the harness the model ran in (Pi, Claude Code, WorkBuddy and others) and the date; the author scores every question with a note, and warns that totals are only comparable within one test version.
- Finding: Gemini 3.8 Flash scored 49/60 on version 2.1 in the DeepSeek harness: perfect constrained sentences and reasoning, but weaker coding, with a space-shooter control bug, a crude racing game with drifting cars and Trello lists that could not be reordered.
- Limitation: One author's scores on six questions; rows use different harnesses and test versions (2.1 or 2.2), and a few rows' table total differs by one point from the author's note.
- Score: 49 / 60 (v2.1)

## Public leaderboard coverage

A board missing from this list has no figure for this model in this snapshot; missing coverage is not a zero.

- LMArena Text Arena — Arena score (Elo): 1495 ±5 (retrieved 2026-10-05) [source](https://arena.ai/leaderboard/text)
- Artificial Analysis Intelligence Index — Intelligence Index: 41 (retrieved 2026-09-21) [source](https://artificialanalysis.ai/leaderboards/models)
- Epoch Capabilities Index (ECI) — General ECI: 157 (155 - 160) (retrieved 2026-10-04) [source](https://epoch.ai/eci)
- Vals Index — Accuracy (GDP-weighted finance, coding, legal and tax): 55.95 ±1.14 (retrieved 2026-10-23) [source](https://www.vals.ai/benchmarks/vals_index)
- LMArena Text Arena · Chinese — Arena score (Elo): 1536 ±14 (retrieved 2026-10-05) [source](https://arena.ai/leaderboard/text/chinese)
- LMArena Text Arena · Coding — Arena score (Elo): 1530 ±8 (retrieved 2026-10-05) [source](https://arena.ai/leaderboard/text/coding)
- Artificial Analysis · Output Speed — Median output tokens/s: 297 (retrieved 2026-09-22) [source](https://artificialanalysis.ai/leaderboards/models)
- Artificial Analysis · GDPval-AA v2.1 — Agentic Real-World Work Tasks, (Elo-500)/2000 (%): 46 (retrieved 2026-09-21) [source](https://artificialanalysis.ai/leaderboards/models)
- Artificial Analysis · Terminal-Bench v4.0 — Agentic Coding & Terminal Use (%): 20 (retrieved 2026-09-21) [source](https://artificialanalysis.ai/leaderboards/models)
- Artificial Analysis · MLCR-AA — Medical Long Context Reasoning (%): 21.7 (retrieved 2026-09-07) [source](https://artificialanalysis.ai/evaluations/mlcr-aa)

## Curated comparisons

- [claude-fable-5-1 vs gemini-3.8-flash](https://dash.chinaapi.ai/models/compare/claude-fable-5-1-vs-gemini-3.8-flash)
- [deepseek-flash vs gemini-3.8-flash](https://dash.chinaapi.ai/models/compare/deepseek-flash-vs-gemini-3.8-flash)
- [gemini-3.8-flash vs glm-5.3-flash](https://dash.chinaapi.ai/models/compare/gemini-3.8-flash-vs-glm-5.3-flash)
- [gemini-3.8-flash vs gpt-6-astra](https://dash.chinaapi.ai/models/compare/gemini-3.8-flash-vs-gpt-6-astra)
- [gemini-3.8-flash vs qwen3.8-flash](https://dash.chinaapi.ai/models/compare/gemini-3.8-flash-vs-qwen3.8-flash)

## Official sources

- [Introducing Gemini 3.8 Flash and 3.8 Flash Cyber](https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/) — Google, retrieved 2026-09-24
- [Gemini 3.8 Flash | Gemini API](https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash) — Google, retrieved 2026-09-24
- [Gemini 3.8 Flash - Model Card](https://deepmind.google/models/model-cards/gemini-3-8-flash/) — Google DeepMind, retrieved 2026-09-24