MODEL VERSIONS ON THIS PAGE HAVE MOVED — CHECKED 8 OCTOBER 2026
Current line, with published API pricing per million tokens:
●
OpenAI — GPT-6 Sol (paid ChatGPT), GPT-6 Luna ($0.10 / $0.50), GPT-6.1 Sol ($2.00 / $10.00), GPT-6 Astra
●
Anthropic — Claude Opus 5.5, Sonnet 5.5 ($2.00 / $10.00), Haiku 5.5 ($0.10 / $0.50 below 100K), Fable 5.1
●
xAI — Grok 4.3 ($1.25 / $2.50), Grok 4.7 ($2.00 / $6.00)
●
Google — Gemini 3.5 Flash-Lite ($0.30 / $2.50), Gemini 3.8 Flash ($0.75 / $3.75 to 31 Dec), Gemini 4 Argon
The comparison below is kept because the reasoning still holds — how these tiers trade price against capability has not changed. The version numbers and prices have.
Current pricing table →
QUICK VERDICT — JULY 22, 2026
● Better overall intelligence: Kimi K3 — AA Index #4 (score 57) vs 3.6 Flash AA Index 50
● Better price: Gemini 3.6 Flash — $1.50/$7.50/M vs K3's $3/$15/M (50% cheaper)
● Faster output: Gemini 3.6 Flash — 304 tok/s
● Better agentic coding (SWE Marathon): Kimi K3 — 42.0% #1 globally
● Best frontend coding: Kimi K3 — Design Arena #1 (1679 Elo)
● Computer Use built in: Gemini 3.6 Flash — K3 requires separate tooling
● Data residency: 3.6 Flash — Google (US). K3 — Moonshot AI (China NI Law applies to hosted API)
● Open weights: K3 by July 27 — enables safe self-hosting on Western infra
Full Comparison Table
| Model | Input /1M | Output /1M | AA Index | SWE Marathon | Speed | Data |
|---|
| Gemini 3.6 Flash | $1.50 | $7.50 | 50 | N/A | 304 tok/s | US ✓ |
| Kimi K3 (Allegretto+) | $3 | $15 | 57 (#4) | 42.0% #1 | N/A | China NI Law ⚠ |
Which to Choose
Gemini 3.6 Flash: High-speed high-volume agentic pipelines, multimodal tasks with Computer Use, teams in the Google ecosystem, regulated industries where Chinese data residency is not acceptable, and workloads where $1.50/M input justifies the intelligence tier below K3.
Kimi K3 (Allegretto+): Frontend development (Design Arena #1), long-horizon agentic coding (SWE Marathon 42.0% #1), higher overall intelligence tasks (AA Index #4 vs 3.6 Flash's 50), and workloads where data residency is acceptable or where you will self-host after July 27.
After July 27: K3 open weights ship. Self-hosting on Google Cloud or AWS Western regions eliminates the China NI Law concern and removes the data residency disadvantage. A self-hosted K3 with 3.6 Flash as a fast routing layer is potentially the most capable and cost-efficient architecture available.
Last updated July 22, 2026. Related: Kimi K3 vs GPT-5.6 Sol → · Gemini 3.6 Flash vs Claude Sonnet 5 → · Kimi K3 full review →