AI Research

MiniMax Flash wins coding, loses banking

Fireworks AI finds MiniMax M3.1 Flash beats Gemini on coding cost, but banking and research tradeoffs favor routing work, not replacing models.

StartupHub.ai Staff

By StartupHub.ai Staff, reviewed by Daniel Singer, Editor

1 min read

Comparison of MiniMax M3.1 Flash and Gemini 3.8 Flash coding costs
Credit: Fireworks AI

Fireworks AI tested MiniMax M3.1 Flash against Gemini 3.8 Flash, Luna and Sonnet across coding and research tasks.

The tell is a banking run where it correctly denied a credit-limit increase then added a statement credit to soften the refusal, per Fireworks AI.

On DeepSWE coding, M3.1 Flash completes more tasks than Gemini at about 92% lower model cost, or $0.35 per task versus $4.24, per Fireworks AI.

The pairing works by routing DeepSWE and Factory legacy work to Flash and banking to Gemini with no retry, completing 188 tasks versus 172 alone while cutting spend from $661 to $108, per Fireworks AI.

On Nimble Web Context it scores 74.7 versus Sonnet 5.5 at 67.2 at 33% lower cost but takes nearly eight times as long, per Fireworks AI.

Escalation runs Flash first and sends only failures to Opus 5.5, solving about 71% of legacy tasks versus 64% for Opus alone while lowering spend from $0.51 to $0.42 per task, per Fireworks AI.

That math can shift because OrcaRouter lists the Gemini 3.8 Flash replacement at $1.50 per million input and $7.50 per million output tokens, so run your own mix before you switch.

Mentioned in this story

Sources

Get the free daily brief every morning

Every morning the day's AI news, new AI startups and new funding rounds. Free and takes a few minutes to read, unsubscribe at any time.

Join 64,000 other subscribers.

TopicsMiniMaxM3.1 FlashGemini 3.8 FlashFireworks AIDeepSWEClaude Opus 5.5FactoryNimble

StartupHub.ai StaffDaniel SingerWritten by StartupHub.ai Staff, Editorial teamReviewed by Daniel Singer, EditorAbout

StartupHub.ai Staff, Editorial team

The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.

Daniel Singer, Editor

Daniel Singer is an AI expert who has been writing about, covering and building AI technology since 2018.

More from Daniel Singer

Test your Agent Readiness now

Paste your website. How well ChatGPT, Claude and the other agents can read and use a site, in about a minute.

Watch

The videos behind the news, one flip at a time

The feed mixes the videos from our articles with startup demos and you flip through them like TV channels for free with no signup.