## What Models Compete with Claude Sonnet 5? Based on the launch data from June 30, 2026 and cross-vendor benchmarks released by Anthropic, here are the **direct competitors** to Claude Sonnet 5: ### 🥇 Direct Competitors | Model | Company | SWE-bench Verified | GPQA Diamond | ARC-AGI-2 | OSWorld-Verified | |---|---|---|---|---|---| | **Claude Sonnet 5** 🏆 | **Anthropic** | **82.1%** | **96.2%** | **84.7%** | **88.3%** | | **Gemini 3.1 Pro** | Google | 80.6% | 94.3% | 77.1% | — | | **GPT-5.4** | OpenAI | ~80% | — | — | 75.0% | | **Claude Sonnet 4.6** | Anthropic (prev. gen) | 79.6% | — | — | 78.5% | ### 💡 What the benchmarks say Claude Sonnet 5 **leads across all major benchmarks** as of its launch date: - **SWE-bench Verified (agentic coding):** First model to break **80%**, scoring 82.1% — ahead of Gemini 3.1 Pro (80.6%) and GPT-5.4 (~80%) - **GPQA Diamond (PhD-level science):** 96.2%, surpassing Gemini 3.1 Pro's prior best of 94.3% - **ARC-AGI-2 (abstract novel reasoning):** 84.7%, a 7-point lead over Gemini 3.1 Pro (77.1%) - **OSWorld-Verified (computer use):** 88.3%, well ahead of GPT-5.4 (75.0%) and the **human expert baseline** (72.4%) ### 💰 Pricing Comparison | Model | Input ($/M tok) | Output ($/M tok) | |---|---|---| | **Sonnet 5** (intro through Aug 31) | **$2** | **$10** | | **Sonnet 5** (standard) | $3 | $15 | | **Sonnet 4.6** | $3 | $15 | | **GPT-5.4** (est.) | $5–$15 | $15–$60 | | **Gemini 3.1 Pro** (est.) | ~$2–$5 | ~$10–$20 | ### 🔑 Summary The two main competitors to Claude Sonnet 5 are **Google's Gemini 3.1 Pro** (the closest rival on SWE-bench and GPQA) and **OpenAI's GPT-5.4** (strongest on coding). Claude Sonnet 5 edges ahead of both on nearly every published metric — and does so at a **competitive price point** ($3/$15 per M tokens standard, with introductory pricing of $2/$10). Internally, it also directly competes with Anthropic's own **Opus 4.8** (stronger overall but costs nearly double), positioning Sonnet 5 as the "best value" mid-tier model that delivers near-flagship performance.