Gemini 3.8 Flash and 3.8 Flash Cyber

Who Should Read

A backend developer running coding agents or long-running automation on the Gemini API who is weighing a model upgrade against cost

Core Mechanics

• 3.8 Flash costs the same as 3.7
  • The introductory price is $0.75 per million input tokens and $3.75 per million output tokens.
  • It lasts until December 31, 2026, and from January 1, 2027 it becomes $1.50 input and $7.50 output.
• Coding and agent performance beat 3.7 Flash
  • On DeepSWE v1.1 (long-horizon software engineering), it reportedly solves complex problems end to end better than most larger frontier models.
  • It also beats 3.7 Flash and other frontier models on Vals Finance Agent V2 and Harvey's Legal Agent Benchmark.
  • It scores 54.9% on HLE-Verified.
• It works harder by spending more tokens
  • On complex tasks it takes extra reasoning steps and calls tools iteratively.
  • Token use can rise, especially at higher effort levels.
• There are options if efficiency matters most
  • Lower effort levels reduce token overhead.
  • 3.7 Flash remains fully supported for efficiency-first workloads.

Community Reactions

How to Apply

• Run your existing tasks on 3.8 Flash and compare cost
  • Run 5 to 10 of your current agent tasks on both 3.7 Flash and 3.8 Flash.
  • Record success rate and input/output token counts for each.
• Measure token use at each effort level
  • Run the same task at low and high effort.
  • Find the point where quality gains justify the extra tokens.
• Route models by task difficulty
  • Send long-horizon coding and multi-step reasoning to 3.8 Flash.
  • Keep simple, high-volume work on 3.7 Flash.
• Put the 2027 price increase in your budget
  • Recalculate monthly cost at $1.50 input and $7.50 output per million tokens, which apply from January 1, 2027.

Try It

# Prompt to paste into Claude or ChatGPT (cost calculation for moving to 3.8 Flash)

I'm deciding whether to move from Gemini 3.7 Flash to 3.8 Flash.
Pricing: introductory (until 2026-12-31) $0.75 per 1M input tokens, $3.75 per 1M output tokens. From 2027-01-01: $1.50 input, $7.50 output.

My monthly usage:
- Input tokens: {{monthly input tokens}}
- Output tokens: {{monthly output tokens}}
- Assume 3.8 Flash uses {{increase}}% more output tokens than 3.7

Instruction: Calculate monthly cost in a table for the introductory period and for 2027 onward, comparing increases of 0%, 20%, and 50%.

Next step: Based on the results, decide what token increase is acceptable and whether to lower effort.

Terminology

Related Papers

• Mercury 3: Diffusion at scale

  Scaling diffusion LLM 40x gives Mercury 3: 4–8x faster than speed-tier models, similar quality

• Timestral: Mistral's time-series foundation models

  A 209M time-series model ranks first on three benchmarks and runs on CPU.

• DeepSeek V4.1 Flash 1017 (Max)

  DeepSeek V4.1 Flash 1017 Max scores 45 on the Intelligence Index at 211 tok/s and $0.30 input.

• If you don't run your own LLM proxy, how do you know whats going on?

  Running your own LLM proxy shows which keys, models, and cache misses drive your usage.

• DeepSeek: DeepSeek v4.1 Flash 1017

  DeepSeek V4.1 Flash 1017: $0.05/M input, 1M context, built for agents

• DeepSeek-v4.1-Flash-1017

  DeepSeek V4.1 Flash minor update lifts coding and agent benchmarks while cutting run time and output length

• Show HN: Llmspeed – feel how fast a local LLM runs before you buy the hardware

Related Resources

• https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/