Matching posts

Matching terms are highlighted

Local DeepSeek V4.1 Flash draws charts by writing Python and self-checking with vision

A user shows local DeepSeek V4.1 Flash generating charts and diagrams by writing Python scripts that render PNGs, then using vision to review and fix the output, all on-device.

deepseek-v4.1@MiaAI_lab　Filed in No.020

20 hours agoCopy original

POST

600M Persian vision model claims benchmarks rival Gemini Argon

A developer ships a 600M-parameter Persian vision model built on Nemotron 3.5 Parse, claiming it competes with Gemini Argon on their benchmarks.

gemini-4@iamRezaSayar　Filed in No.008

20 hours agoCopy original

Olaolu Ariyo's visual artist site in Birmingham

Birmingham-based visual artist Olaolu Ariyo's personal site showcasing his artwork with a distinct, expressive style.

个人网站@olaoluariyo　Filed in No.005

2 days agoCopy original

GPT-6.1 Terra launches: from visual understanding to real操作

OpenAI releases the fast, low-cost Terra, up 15 points on a PC-operation benchmark over GPT-5.6 Terra and near Astra performance at roughly a tenth of the token price.

gpt-6@itnavi2022　Filed in No.001

1 day agoCopy original

Time-limited DGX Spark agent benchmark: MiniMax M3.1 Flash scores 26/44 with best vision

On 44 auto-graded time-limited agent tasks, DeepSeek V4.1 Flash scored 40/44, GLM-5.3 37/44, GLM-5.3-Flash 33/44, and MiniMax M3.1 Flash 26/44 but with the best vision; low effort beat max for DeepSeek and GLM.

minimax-m3.1@majewskizby　Filed in No.021

3 days agoCopy original

User calls DeepSeek V4.1 Flash 1017 an S-tier model for 3D and visual work

A user says V4.1 Flash 1017 may be S-tier, praising its reasoning style, fast 3D/visual performance and $0.10 price.

deepseek-v4.1@shamanic_arts　Filed in No.020

2 days agoCopy original

POST

Voice-prompt app building workflow: OpenCode with DeepSeek V4.1 Flash and GLM 5.3

The author describes a voice-prompt plus visual-reference app building loop, mixing Claude Code with Opus 5.5 and OpenCode with DeepSeek V4.1 Flash / GLM 5.3 as harnesses.

deepseek-v4.1@nutlope　Filed in No.020

2 days agoCopy original

Lovart 2.0 review: one prompt builds a full personal brand kit

A user generated a logo, avatar, X header, Telegram cover and post templates from a single Lovart 2.0 prompt, praising Arti for remembering their style and flagging design issues.

lovart@arthurcollinsai　Filed in No.017

3 days agoCopy original

Codex with GPT-6 Astra for websites: better visuals, weaker detail than Opus 5.5

The author built several sites with Codex and GPT-6 Astra, finding visuals and hand-drawn logos stronger but translation skipping passages and coding detail still behind Claude Opus 5.5.

gpt-6@KenLin1985　Filed in No.001

3 days agoCopy original

Doodle recognition showdown: M3.1 Flash vs DeepSeek V4.1 Flash vs Nemotron VL

A demo game has three vision models, including MiniMax M3.1 Flash and DeepSeek V4.1 Flash, race to recognize doodles as they are drawn, all served on Nebius.

minimax-m3.1@sujee_dev　Filed in No.021

4 days agoCopy original

Showing10of60posts

10 per batch

Keep scrolling to reveal posts 11–20

Showing the first 60 matches. Narrow the query with a model name or tag to find more specific results.