Gemini 3.8 Flash

AGENTIC WORKHORSE

Released Sept 2, 2026; Google DeepMind's frontier workhorse for agentic coding loops, recursive self-correction, and 1M token real-time multimodal streaming.

Context

1,048,576 tokens

Architecture

TPU v6e Speculative MoE

Official API (1M)

$0.75 in / $3.75 out

License

Google AI Studio / Vertex AI

Published benchmarks · lab-reported

Vendor-published figures, not independently reproduced. Sources in Verification below.

SWE-BENCH

76.4%

MMLU-PRO

86.8%

GPQA DIAMOND

73.2%

Quickstart

curl -s https://modelregistry.tirup.in/api/cli?model=gemini-3-8-flash

OFFICIAL ANNOUNCEMENT API DOCS

BACK TO LEDGER

Sources

Gemini 3.8 Flash announcement · Google DeepMind
Gemini 3.8 Flash API reference · Google DeepMind
Gemini 3.8 Flash model card · Google DeepMind

Report an error in this record