ChatGPT vs Claude vs Gemini vs Grok

OpenAI's ChatGPT remains the most widely-used AI assistant in the world. GPT-5.5 (April 23, 2026) held #1 on the Artificial Analysis Intelligence Index at 60.2 until Anthropic retook the crown, and after the June 9 release of Claude Fable 5 (64.9) it now sits behind both Fable 5 and Claude Opus 4.8 (61.4), still the strongest non-Claude model. It dominates shell automation (Terminal-Bench 2.0: 82.7%, +13 over Opus 4.7) and advanced math (FrontierMath Tier 4: 35.4%). API pricing doubled vs 5.4 to $5/$30 per million tokens; a new GPT-5.5 Pro variant for longer reasoning sits at $30/$180. Codex now has a 1M-token context window with optional fast-mode at 2.5x cost. Honest caveats: GPT-5.5 still loses SWE-Bench Pro to Claude Opus 4.8 (58.6% vs 69.2%), loses MCP-Atlas tool use to both Opus (82.2%) and Gemini (83.6%), and posts an 86% hallucination rate on AA-Omniscience. The ecosystem advantage remains unmatched: DALL-E, Codex, Atlas browser, 60+ connectors, Memory, Projects, GPT Store, and Microsoft 365 Copilot integration. Sora video app/API is being discontinued (web/app April 26, 2026; API September 24, 2026).

Anthropic's Claude Fable 5 (June 9, 2026) is the most capable model the company has ever made generally available, and it tops the Artificial Analysis Intelligence Index at 64.9, nearly five points clear of any other lab's best model and ahead of Anthropic's own Opus 4.8 (61.4) and GPT-5.5 (60.2). It is the production, safeguarded version of the same weights as the restricted Claude Mythos 5. On coding it posts 95.0% on SWE-bench Verified and 80.3% on SWE-bench Pro, beating Opus 4.8 (69.2%), GPT-5.5 (58.6%), and Gemini (54.2%), and on GDPval-AA, the benchmark for real economic-value work, it leads at 1,932 Elo. It is built for long-horizon autonomy, working for days at a time in an agent harness and testing its own output, plus state-of-the-art vision for diagrams, charts, and tables inside PDFs. New wrinkle for developers: Fable 5 ships safety classifiers that can decline a request (returned as stop_reason "refusal", with server, client, or manual fallback to Opus 4.8), reroute in under 5% of sessions, and require 30-day data retention. Fable 5 prices at $10/$50 per million tokens with a 1M-token context and 128K output. Sitting alongside it, the cheaper, faster Claude Opus 4.8 ($5/$25, optional fast mode at $10/$50) remains the everyday workhorse. Honest caveats: Fable runs slower per turn, costs double Opus 4.8, and its classifiers have refused some innocuous prompts near security and biology topics.

Google's Gemini 3.5 Flash (May 19, 2026) is the new headline model: a Flash-tier model that beats last generation's flagship 3.1 Pro on coding and agentic work. It posts 76.2% on Terminal-Bench 2.1, 83.6% on MCP Atlas tool-use, and 1656 Elo on GDPval-AA, while running ~280 tokens/sec (one of the fastest models measured) at $1.50/$9.00 per million tokens. On the Artificial Analysis Intelligence Index it debuted at 55, behind Claude Fable 5 (64.9), Claude Opus 4.8 (61.4), and GPT-5.5 (60.2) but sitting on the speed-intelligence frontier, though the newer Grok 4.5 has since edged ahead of Gemini on the current index. Honest caveat: 3.1 Pro still wins some of the hardest abstract-reasoning tests like ARC-AGI-2, and 3.5 Flash costs 3x the previous Gemini 3 Flash. Native multimodal across text, image, audio, and video with 1M-token context. Gemini 3.5 Pro is slated for next month. Deep integration across Google Workspace, NotebookLM, Veo 3.1, and the new Google Antigravity 2.0 agent platform.

xAI's Grok 4.5 (July 8, 2026) is the new flagship, pitched by Elon Musk as "an Opus-class model, but faster." Released under the SpaceXAI banner (xAI merged with SpaceX in February 2026, and a $60B acquisition of Cursor-maker Anysphere is signed, expected to close in Q3), it was trained alongside Cursor on tens of thousands of NVIDIA GB300 GPUs and is reportedly 1.5T parameters, 3x its predecessor. It ranks 4th of 168 on the Artificial Analysis Intelligence Index at 54, behind Claude Fable 5 (64.9), Claude Opus 4.8 (61.4), and GPT-5.5 (60.2) but now ahead of Google's Gemini models on the current index, xAI's first genuinely frontier-adjacent showing. It also won Snorkel's independent GDPVal+ professional-work benchmark (29% mean pass rate vs GPT-5.5's 22% and Opus 4.8's 21%), though the test harnesses differed per model. The real story is economics: $2/$6 per million tokens (cached input $0.50), roughly half of comparable frontier pricing, with vendor-claimed 2x token efficiency (15,954 average output tokens per SWE-Bench Pro task vs 67,020 for Opus 4.8 at max reasoning) and ~80-90 tokens/sec serving speed. Honest caveats: the context window stepped down to 500K tokens, half of Grok 4.3's 1M (API rates double past 200K input, and xAI says a 1M upgrade is imminent); on xAI's own published coding benchmarks it trails Claude Fable 5 on three of four (SWE-Bench Pro 64.7% vs Fable's 80.4%); on the neutral-harness DeepSWE 1.1 it falls behind Opus 4.8, GPT-5.5, and Fable 5; and EU availability only lands mid-July. Grok 4.3 stays in the lineup as the budget 1M-context option at $1.25/$2.50.