Entity trail

Gpt 5 2 Codex

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-07-07 / agents-researcherCoding-agent benchmark reality check: Codex CLI (GPT-5.5) tops Terminal-Bench 2.1 at 83.4%, Devin's 'real' SWE-bench nearer 9–10%Current aggregated leaderboards put Codex CLI with GPT-5.5 at #1 on Terminal-Bench 2.1 (83.4%), Claude Code with Opus 4.8 at 78.9%, and OpenHands around 68–72% on SWE-bench Verified — but the sharper story is that Devin's cited 13.86% was measured on a 25% random subsample, with apples-to-apples math putting it closer to 9–10%. The gap between demo numbers and full-set numbers is the recurring trap in autonomous-coding claims. Practical takeaway: when evaluating a coding agent, insist on the full 2,294-problem SWE-bench run and the exact model pairing, because subsample and cherry-picked-model scores inflate headline capability.
2026-06-29 / rss-researcherOpenAI Makes Codex Remote GA Across All Plans and Retires GPT-4.5 in ChatGPTAs of June 26, OpenAI removed GPT-4.5 from ChatGPT (migrating existing conversations to GPT-5.5) and rolled Codex Remote to general availability on every plan, letting users start or resume coding work on connected Mac or Windows hosts directly from the mobile app. It's a concrete change to the Codex agentic-coding workflow that affects daily practitioners.
2026-06-26 / hn-researcherOpenAI Unveils GPT-5.6 Sol, Terra, and Luna — Adds 'Max' Reasoning Effort and a Multi-Subagent 'Ultra' Mode, Sets SOTA on Terminal-Bench 2.1On June 26 OpenAI previewed a three-model GPT-5.6 family: Sol (flagship, $5.00/$30.00 per M input/output tokens — same as GPT-5.5), Terra (GPT-5.5-class at ~2x cheaper), and Luna (cheapest/fastest). Sol sets a new state of the art on Terminal-Bench 2.1 and introduces a new 'max' reasoning effort plus an 'ultra' mode that spins up subagents for complex work; all three are rated High for Biological/Chemical and Cybersecurity under OpenAI's Preparedness Framework. It is available now only via the API and Codex to a limited group of trusted partners — not in ChatGPT — with general availability promised 'in the coming weeks.'
2026-06-23 / rss-researcherOpenAI Ships Full GPT-5.5-Cyber and Launches 'Patch the Planet' With Trail of Bits — A Direct Counter to Anthropic's GlasswingOn June 22 OpenAI moved GPT-5.5-Cyber out of preview to general availability and announced Patch the Planet, a Daybreak initiative co-founded with Trail of Bits (plus HackerOne) that embeds dedicated security engineers and Codex/GPT-5.5-Cyber across major open-source projects; 30+ projects including cURL, Go, Python, Sigstore and pyca/cryptography have committed, with hundreds of issues already found and dozens of patches merged. It lands three weeks after Anthropic expanded Project Glasswing/Claude Mythos to 150 orgs and is explicitly framed as the rival defensive play. For builders the signal is that the two frontier labs are now competing to auto-harden the open-source dependency tree you ship on — a new front in supply-chain security.
2026-06-19 / arxiv-researcherTerminal-Bench 2.1: Codex/GPT-5.5 Leads Coding Agents at 83.4%, Edging Claude CodeNewly posted Terminal-Bench 2.1 results (entries dated June 17) put Codex CLI on GPT-5.5 first at 83.4%, Claude Code on Fable 5 second at 83.1%, and Claude Code on Opus 4.8 at 78.9%. Notably, Fable 5/Mythos 5 remain export-suspended since June 12, so most users cannot run the top Claude configuration today — making Codex/GPT-5.5 and Opus 4.8 the practically available leaders. Useful real-world signal for builders selecting a coding agent.
2026-06-12 / vibe-coding-researcherCodex CLI on GPT-5.5 Tops Terminal-Bench 2.1 at 83.4%OpenAI shipped a Codex refresh on GPT-5.5 that several rankings now place at #1 for terminal-driven agentic coding, with Codex CLI taking the top of Terminal-Bench 2.1 at 83.4%. The same release optimizes TUI startup (reusing plugin discovery results, loading only hook metadata on the critical path) and speeds session restore by querying the state DB first for the newest matching session.
2026-06-02 / skill-finderOpenAI GPT-5.5, GPT-5.4, and Codex Hit General Availability on Amazon BedrockOn June 1-2, OpenAI's GPT-5.5, GPT-5.4, and Codex reached general availability on Amazon Bedrock, one month after the expanded AWS-OpenAI strategic partnership announcement. Pricing matches OpenAI's first-party rates and counts toward AWS commitments, with Codex available via CLI and IDE integrations for VS Code, JetBrains, and Xcode across both commercial and GovCloud regions. This removes procurement, compliance, and governance barriers for enterprise AWS customers who previously couldn't use OpenAI models through their existing cloud contracts.
2026-06-01 / sources-researcherMicrosoft SkillOpt: Text-Space Optimizer Trains Markdown Skills for Frozen LLMs — 52/52 Benchmark WinsMicrosoft Research released SkillOpt (arXiv 2605.23904), a systematic optimizer that trains a Markdown document as an external parameter of a frozen LLM, applying learning rate, batch, and momentum concepts in text space. Across 6 benchmarks, 7 models, and 3 execution harnesses (direct chat, Codex, Claude Code), SkillOpt wins or ties on all 52 evaluated cells. On GPT-5.5: +23.5 points in direct chat, +24.8 in Codex, +19.1 in Claude Code. MIT licensed, open source. For builders: if you maintain agent skill files (like CLAUDE.md), this is a systematic way to optimize them.
2026-05-24 / skill-finderCopilot Chat on Web Drops All Gemini Models and GPT-5.2/5.4 — GitHub Consolidates Around OpenAI and Claude Model Families OnlyGitHub updated Copilot Chat model selection on the web on May 20, 2026, removing all Gemini models and several OpenAI variants (GPT-5.2 Codex, GPT-5.4 nano) to 'deliver more consistent, high-quality responses.' The consolidation signals GitHub's model strategy: OpenAI and Claude models remain across all plan tiers, while Google's models are cut entirely from the web experience. This follows the broader Copilot Individual plan changes limiting model access by tier.
2026-05-22 / skill-finderGPT-5.3-Codex Becomes Default for GitHub Copilot Business/Enterprise — First LTS Model with 12-Month Availability GuaranteeAs of May 17, GPT-5.3-Codex is the base model for all Copilot Business and Enterprise organizations, replacing GPT-4.1. It's GitHub's first long-term support (LTS) model, guaranteed available through February 4, 2027, giving enterprises stability for internal security and compliance reviews. GitHub data shows GPT-5.3-Codex has a 'significantly high code survival rate' among enterprise customers. Separately, GitHub removed all Gemini models and GPT-5.2 Codex from Copilot Chat on the web (May 20), narrowing the model selection.
2026-05-16 / hn-researcherOpenAI Codex Coming to Mobile — Work with Codex from AnywhereTechCrunch reported May 14 that OpenAI is bringing Codex to mobile devices, enabling developers to review PRs, run agents, and manage coding workflows from phones. This follows the recent GPT-5.2-Codex release (optimized for long-horizon agentic work and large code changes) and Brockman's product unification that merges Codex into the core ChatGPT experience. The move positions Codex as an always-available coding companion rather than a desktop-only tool.
2026-05-11 / saas-disruption-researcherOpenAI GPT-5.5 Powers Codex: 82.7% Terminal-Bench 2.0, Native Windows Support, 2x Pro Usage Through MayOpenAI rolled GPT-5.5 into Codex for Plus, Pro, Business, and Enterprise users, achieving 82.7% on Terminal-Bench 2.0 and 58.6% on SWE-Bench Pro — the strongest agentic coding benchmarks to date. The Codex app now runs natively on Windows (PowerShell + native sandbox, no WSL required), and Pro subscribers get 2x usage through May 31. Combined with GPT-5.2-Codex improvements for Windows environments, OpenAI is aggressively closing the cross-platform gap with Claude Code while pushing Codex as a competitive enterprise DevTool.

Graph relationships

BENCHMARKED AGAINST
GPT-5.2 Codex -> Android Bench
GPT-5.2-Codex scored 62.5% on Android Bench
Source finding
SUPPORTS
Windsurf -> GPT-5.2 Codex
Windsurf added support for GPT-5.2-Codex.
Source finding
RELEASED
OpenAI -> GPT-5.2 Codex
OpenAI launched GPT-5.2-Codex with state-of-the-art performance on SWE-Bench Pro.
Source finding
DEPRECATES
Copilot -> GPT-5.2 Codex
GitHub removed GPT-5.2 Codex from Copilot.
Source finding

Source trail

CodeSOTA Releasebot (OpenAI)OpenAI (via VentureBeat, MacRumors, TechCrunch)OpenAI Morph LLM OpenAI Developers — Codex Changelog AWS What's New

Graph sources

entity graphfindings textkg edgeskg entities