# Gemini 3.8 Flash vs Kimi K3 (2026): side-by-side comparison Source: [GLAD-AI-TOR](https://glad-ia-tor.com) · Full page: https://glad-ia-tor.com/vs/gemini-3-8-flash-vs-kimi-k3 Arena: llm-models · Crowd scores are live visitor verdicts (one per person per tool, never paid, Bayesian-smoothed). ## At a glance | | Gemini 3.8 Flash | Kimi K3 | |---|---|---| | Price | $3.75/1M out | $15/1M out | | Crowd score | 60% (1 votes) | 57% (3 votes) | | provider | Google | Moonshot AI | | contextWindow | 1M tokens | 1M tokens | | priceIn | $0.75/1M in | $3/1M in | | priceOut | $3.75/1M out | $15/1M out | | modalities | text, vision, audio, video, PDF | text, vision, video input | | openWeights | no | yes | | reasoning (1-5) | 4 | 4.5 | | coding (1-5) | 4 | 4.5 | | writing (1-5) | 4 | 4 | | speed (1-5) | 5 | 2 | | valueForMoney (1-5) | 5 | 3.5 | ### Gemini 3.8 Flash > Google workhorse September 2026: 1M context, $0.75/$3.75 promo pricing, 305 tokens/s output speed Strengths: - Fastest output speed measured by Artificial Analysis at 305 tokens/s, ideal for interactive applications - Promo pricing $0.75/$3.75 through end of 2026 is roughly 1/13 the cost of Claude Fable 5.1 - 89.4% on Terminal-bench 2.1 vs 85.8% for 3.7 Flash, beats Claude Opus 5 on three Google benchmarks - Full multimodal: text, image, video, audio, and PDF input in a single model Weaknesses: - Promo pricing doubles to $1.50/$7.50 on January 1, 2027, plan budgets accordingly - Built on 3.7 Flash architecture rather than a new base model, incremental rather than generational - Proprietary model under Google product and API terms Verdict: Gemini 3.8 Flash is the budget king for 2026. At promo pricing it costs roughly 1/13 of Fable 5.1 while matching or beating it on several coding benchmarks, and 305 tokens/s output speed makes it the fastest frontier-tier model available. Choose it for high-volume pipelines, interactive products, and any workflow where latency matters more than peak reasoning. The catch: promo ends December 31, so build your 2027 budget at $1.50/$7.50. For pure reasoning ceiling, Fable 5.1 and GPT-6 Astra remain ahead. Full review: https://glad-ia-tor.com/tool/gemini-3-8-flash · Markdown: https://glad-ia-tor.com/tool/gemini-3-8-flash.md ### Kimi K3 > Moonshot's 2.8T-parameter MoE that sparked the 'new DeepSeek moment': frontier-adjacent scores at $3/$15 Strengths: - #3 on the Artificial Analysis Intelligence Index (57) at launch, comparable to Claude Opus 4.8 and GPT-5.5: the closest a Chinese lab has come to the closed US frontier - 93.4% on SWE-bench Verified in Vals AI's independent harness (GPT-5.6 Sol: 96.2%, Claude Fable 5: 95.0%) and #1 on Arena.ai's Frontend Code Arena at 1679 points, ahead of Fable 5 - 93.5% on GPQA Diamond (between Fable 5's 92.6% and GPT-5.6's 94.1%), 96.1% on AIME 2025, and Elo 1668 on GDPval-AA v2 agentic work, second only to Fable 5 - Strong agentic profile: #1 on AutomationBench-AA SaaS workflows (53%), long-horizon terminal and repo navigation via Kimi Code, and ~21% fewer output tokens than K2 for more intelligence Weaknesses: - 3x price jump over Kimi K2.6 ($0.95/$4 to $3/$15) makes it the most expensive Chinese model ever shipped, at Claude Sonnet 5 list price: the '10x cheaper than US models' era is over (Simon Willison documented the hike) - Slow: 33 output tokens/s, ranked #145 of 190 models on Artificial Analysis, a poor fit for interactive use - Hallucination rate climbed from 39% (K2.6) to 51% on AA-Omniscience as the model now attempts answers it would previously refuse (Fable 5 sits at a comparable 54.9%) Verdict: Kimi K3 is the strongest argument yet that the frontier is no longer exclusively American: #3 on Artificial Analysis, top-tier GPQA and SWE-bench Verified scores, and the best frontend-code arena ranking in the business, at roughly half to a third of US closed-frontier prices. Take it for agentic coding, frontend work and SaaS automation where its benchmarks are strongest, or if your roadmap depends on self-hosting a frontier-class model once the weights land. Skip it for interactive products (33 tokens/s is slow), for anything touching politically sensitive content or strict EU data-residency requirements (Mandarin-language censorship, Beijing jurisdiction), and for high-accuracy retrieval where its 51% hallucination rate on AA-Omniscience demands a verification layer. Cost-obsessed teams should note DeepSeek V4 still delivers vastly more tokens per dollar; K3's pitch is peak capability per dollar, not cheapest tokens. Full review: https://glad-ia-tor.com/tool/kimi-k3 · Markdown: https://glad-ia-tor.com/tool/kimi-k3.md ## More Full llm-models ranking: https://glad-ia-tor.com/hall-of-fame/llm-models --- This markdown version exists for AI assistants; the canonical page is https://glad-ia-tor.com/vs/gemini-3-8-flash-vs-kimi-k3