Models2026-05-11 · source-backed

GPT-5.5 powers Codex: 82.7% Terminal-Bench 2.0, native Windows, 2x Pro usage through May.

Issue

2026-05-11

Confidence

source-backed

Sources

Redaction

redacted

Story

OpenAI rolled GPT-5.5 into Codex for all paid tiers with 58.6% on SWE-Bench Pro. Codex now runs natively on Windows with PowerShell support, no WSL required. Pro subscribers get doubled usage through May 31.

Codex released Windows / Shared entities / Same source / Earlier coverage / Tension
GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.
Linked by a graph relationship (Codex released Windows); both cover Bench, Codex, GPT, OpenAI; cite the same source (OpenAI).
Codex released Windows / Shared entities / What happened next
GLM 5.2 and the coming AI margin collapse
Linked by a graph relationship (Codex released Windows); both cover Bench, Bench Pro, Codex, GPT; picks up the Bench thread on 2026-07-08.
Codex released Windows / Shared entities / Shared topic / What happened next
Codex CLI on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%.
Linked by a graph relationship (Codex released Windows); both cover Bench, Codex, GPT, OpenAI; overlapping topics (codex, gpt-5).
Claude Code criticizes Windows / Shared entities / Shared topic / What happened next / Tension
Codex on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%, but the top Claude config is export-banned.
Linked by a graph relationship (Claude Code criticizes Windows); both cover Bench, Codex, GPT, Terminal; overlapping topics (codex, gpt-5).
Codex released Windows / Shared entities / Shared topic / What happened next / Tension
Agents' Last Exam: frontier agents pass 2.6% of the hardest economically-valuable tasks.
Linked by a graph relationship (Codex released Windows); both cover Bench, Codex, GPT, Terminal; overlapping topics (codex, gpt-5).
Anthropic partners with OpenAI / Shared entities / Earlier coverage / Tension
Kimi K2.6: A 1-Trillion-Parameter Open-Weights Model Just Beat GPT-5.4 on SWE-Bench Pro
Linked by a graph relationship (Anthropic partners with OpenAI); both cover Bench, Bench Pro, GPT, SWE; earlier Bench coverage from 2026-04-21.
Cursor benchmarked against Codex / Shared entities / What happened next
SpaceXAI and Cursor shipped Grok 4.5, and the number that matters isn't the benchmark rank
Linked by a graph relationship (Cursor benchmarked against Codex); both cover Bench, Bench Pro, GPT, SWE; picks up the Bench thread on 2026-07-13.
Codex released Windows / Shared entities / Same source domain / Shared topic / Earlier coverage / Tension
GPT-5.4 Launches with Native Computer Use and 1M Context.
Linked by a graph relationship (Codex released Windows); both cover Codex, GPT, OpenAI; reported by the same outlet (openai.com).

Source trail

OpenAI

Entities

Bench Bench Pro Codex GPT OpenAI PowerShell SWE Terminal Windows WSL

Provenance

Canonical issue: Ramsay Research Agent — 2026-05-11
AI generated: no
Story unit: 2026-05-11-gpt-5-5-powers-codex-82-7-terminal-bench-2-0-native-windows-2x-pro-usage-through-may
Labels: source-backed, canonical briefing excerpt

Fetching from the wire…

GPT-5.5 powers Codex: 82.7% Terminal-Bench 2.0, native Windows, 2x Pro usage through May.

Story

Related stories

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

GLM 5.2 and the coming AI margin collapse

Codex CLI on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%.

Codex on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%, but the top Claude config is export-banned.

Agents' Last Exam: frontier agents pass 2.6% of the hardest economically-valuable tasks.

Kimi K2.6: A 1-Trillion-Parameter Open-Weights Model Just Beat GPT-5.4 on SWE-Bench Pro

SpaceXAI and Cursor shipped Grok 4.5, and the number that matters isn't the benchmark rank

GPT-5.4 Launches with Native Computer Use and 1M Context.

Source trail

Entities

Provenance

GPT-5.5 powers Codex: 82.7% Terminal-Bench 2.0, native Windows, 2x Pro usage through May.

Story

Related stories

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

GLM 5.2 and the coming AI margin collapse

Codex CLI on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%.

Codex on GPT-5.5 tops Terminal-Bench 2.1 at 83.4%, but the top Claude config is export-banned.

Agents' Last Exam: frontier agents pass 2.6% of the hardest economically-valuable tasks.

Kimi K2.6: A 1-Trillion-Parameter Open-Weights Model Just Beat GPT-5.4 on SWE-Bench Pro

SpaceXAI and Cursor shipped Grok 4.5, and the number that matters isn't the benchmark rank

GPT-5.4 Launches with Native Computer Use and 1M Context.

Source trail

Entities

Provenance