First screen, captured 9 days ago. Open the video.

Images 2.6

What HN thinks

OpenAI drew the rest of the owl; HN counted its legs and checked the invoice.

Still looks like AI slop
Many found the showcase images, including the stained glass whale and GIFs, full of artifacts, yellow tint and physics errors, and debated whether average users even notice or care.
Hidden price hike, mixed quality
Testers found per-token prices unchanged but token usage 3-4x higher at low and medium quality, and some found 2.5 better than 2.6 for design work.
Image gen progress has plateaued
Some argued image models stalled compared to coding models, while others pointed to Flux, Nano Banana Pro and better prompt adherence as real progress.

Consensus: Most agree the new GIF and SVG features are novel but rough, with SVG relying on raster-to-vector conversion and GIFs showing visible splotches.

Recommended in the comments

Nano Banana Pro (service): Cited as best for photos and consistent edits, with near-invisible watermarking.
Recraft ↗ (service): Mentioned as an early native text-to-SVG generator.
Seedream (service): Praised for character consistency.
GenAI Showdown ↗ (service): Shows how open and closed models have progressed since SDXL.
Flux (library): Called a game changer for natural-language prompting in open models.

Summarised by Claude from the comment thread 8 days ago.

Points: 368
Comments: 220
Best rank: #2
Posted: 2026-11-04 by davidbarker
Category: AI & ML

openai gpt-image-2-6 image-generation chatgpt openai-api gif-animation

All captures:

Full page

Captured 9 days ago. Click for the whole screenshot.

Looks similar

Screenshot of coalton-lang.github.io: A Preview of Coalton 0.2

A Preview of Coalton 0.2 (coalton-lang.github.io)

35 points by varjag 8 months ago | 2 comments

Screenshot of hamza.se: Tasfer: a text editor drawn on a 2D canvas

Tasfer: a text editor drawn on a 2D canvas (hamza.se)

84 points by hamza512b 10 days ago | 50 comments

Blog post introducing Tasfer, an open-source local-first text editor that renders everything on an HTML5 2D canvas instead of using contentEditable, with CRDT peer-to-peer sync.

tasfer html5-canvas contenteditable text-editor

Screenshot of sbcl.org: Steel Bank Common Lisp 2.3.8 released: “a mark-region parallel GC is available”

Steel Bank Common Lisp 2.3.8 released: “a mark-region parallel GC is available”(sbcl.org)

140 points by mepian 3 years ago | 12 comments

Screenshot of interdb.jp: The Internals of PostgreSQL

The Internals of PostgreSQL (interdb.jp)

364 points by ibobev 10 days ago | 32 comments

Hironobu Suzuki's free online book explaining PostgreSQL internals, covering storage, query processing, concurrency, VACUUM, buffer manager, WAL, and replication.

postgresql database-internals wal replication

Screenshot of erlang.org: Erlang/OTP 26 Highlights

Erlang/OTP 26 Highlights (erlang.org)

229 points by ramchip 3 years ago | 79 comments

Screenshot of linebender.org: JIT bitmap text rendering for microcontrollers

JIT bitmap text rendering for microcontrollers (linebender.org)

102 points by dmit 5 days ago | 22 comments

Raph Levien's Linebender blog post describing a JIT technique for fast proportional bitmap text rendering on the RP2350 microcontroller in his vic-2350 project.

rp2350 microcontrollers jit bitmap-fonts