LOADING DATASHEET
GROUNDED IN REAL HUMAN CONVERSATIONSYOUR BENCH
GLM 4.7 Flash
◆ OPEN
by Zhipu
OPEN WEIGHTSLOCALAPILONG CONTEXTPAIDREASONING
RANKED #37 OF 316 ASSISTANTS · OVERALL #54 OF 493 RANKED · VIBE SCORE 6.5 · 832 VOICES ·
COPY BADGE pastes a README snippet with the rank SVG.
As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency. It is further optimized for agentic coding use cases, strengthening coding capabilities, long-horizon task planning,...
HAPPY9.3
71 mentions
SAD5.7
60 mentions
VIBE TRAJECTORY
26 weeks · 620 voices in this window
HOLDING UP
Recent vibe is steady. No clear fade in the crowd.
HOW TO USE IT
Open weights. You can use a hosted service, or download it and run it yourself, free.
RUN IT ON YOUR OWN MACHINE
Open weights: download GLM 4.7 Flash once and it's yours. No subscription, no rate limits, works offline.
30B total parameters (MoE with ~3B active), a Q4 file is about 18 GB
- BARE MINIMUM
NVIDIA GeForce RTX 4060 Ti 16GB
16 GB VRAM + 32 GB RAM
~$450 NEWQ4_K_M with partial offload or tight context, ~25-35 tok/s at 8k context
- SWEET SPOT
NVIDIA GeForce RTX 3090 24GB
24 GB VRAM
~$750 USEDQ4_K_XL or Q5_K_M fully in VRAM, ~60-80 tok/s at 32k context
- FULL POWER
Apple Mac Studio M2 Ultra (64GB)
64 GB UNIFIED MEMORY
~$2600 USEDQ8 / BF16 unquantized weights, full context handling up to 128k
LEARN HOW TO SET IT UP
WILL MY BOX RUN GLM 4.7 FLASH?
The three cards above are researched picks. Search the machine you actually have — we only say yes if it should reply at a usable speed, not just load the weights and crawl.
HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.
WHAT CAN YOU ACTUALLY DO WITH IT?
Everyday jobs on each of those rigs: how long GLM 4.7 Flash takes, and how good the crowd says it is at that kind of work.
| THE JOB | BARE MINIMUMNVIDIA GeForce RTX 4060 Ti 16GB | SWEET SPOTNVIDIA GeForce RTX 3090 24GB | FULL POWERApple Mac Studio M2 Ultra (64GB) |
|---|---|---|---|
| Write an emaila paragraph or two, drafted from a one-line brief | ~17 S★★★★★ | ~7 S★★★★★ | —★★★★★ |
| Summarize a documenta long report boiled down to the points that matter | ~50 S★★★★★ | ~21 S★★★★★ | —★★★★★ |
| Build a websitea small landing page, markup and styles together | ~4.4 MIN★★★★★ | ~1.9 MIN★★★★★ | —★★★★★ |
| Build a backendan API with routes, storage and tests | ~14 MIN★★★★★ | ~6 MIN★★★★★ | —★★★★★ |
| Build a gamea playable browser game in one file | ~8.3 MIN★★★★★ | ~3.6 MIN★★★★★ | —★★★★★ |
STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. A DASH MEANS NO SPEED WAS RESEARCHED FOR THAT RIG. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.
WHAT THE INTERNET SAYS
People are mostly frustrated with GLM 4.7 Flash right now. Biggest praise: help with code. Biggest gripe: saying no too much.
627 POSTS · 205 COMMENTS · REDDIT · LEMMY · GITHUB · HACKER NEWS · X · DEV FORUMS · BLUESKY · BLOG · DEV.TO
PEOPLE LOVE
- People feel it is worth the money (4 mentions)
PEOPLE COMPLAIN ABOUT
- Refuses or lectures more than people like (9 mentions)
- Sometimes makes things up (8 mentions)
- Coding help disappoints some people (5 mentions)
- Outages, errors, or limits get in the way (3 mentions)
TOPIC BY TOPIC
- Getting facts rightA SORE SPOT
3 thumbs up · 8 thumbs down
- Help with codeMIXED FEELINGS
6 thumbs up · 5 thumbs down
- Saying no too muchA SORE SPOT
0 thumbs up · 9 thumbs down
- PriceMOSTLY POSITIVE
4 thumbs up · 3 thumbs down
- Working when you need itA SORE SPOT
0 thumbs up · 3 thumbs down
LOOKING FOR AN ALTERNATIVE?
Other models the crowd has fully reviewed, starting with text models like this one.
Muse Spark 1.3
DeepSeek V4 Flash> SOURCES (12)
- REDDIT · 2026-10-19 · 645 UPVOTESSix frontier AIs competed to design and 3D-print egg-protecting capsules. Claude Opus 5.5's egg survived a 20-foot drop without a crack.…
Six frontier AIs competed to design and 3D-print egg-protecting capsules. Claude Opus 5.5's egg survived a 20-foot drop without a crack. GPT-6 Astra failed at 8 feet, while Gemini 4 Argon, Grok 4.7, Qwen 3.8-Max, and Muse Wonder failed at 4 feet. Claude…
- REDDIT · 2026-10-11 · 612 UPVOTESEvery Model That Can Be Run On 10-16GB VRAM Ranked
Every Model That Can Be Run On 10-16GB VRAM Ranked
- REDDIT · 2026-02-17 · 587 UPVOTESCar Wash Test on 53 leading models: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?”
Car Wash Test on 53 leading models: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?”
- REDDIT · 2026-10-19 · 447 UPVOTESA writeup on Gemini 4 Argon’s filters, hidden injected instructions, and how to bypass the injection
A writeup on Gemini 4 Argon’s filters, hidden injected instructions, and how to bypass the injection
- REDDIT · 2026-01-19 · 328 UPVOTEShas anyone tried Claude Code with local model? Ollama just drop an official support
has anyone tried Claude Code with local model? Ollama just drop an official support
- REDDIT · 2026-07-24 · 264 UPVOTESFreaky Frankenstein 5: Internal States. Beta Round 2: Final Testing. Fable 5 Uncensored- Positivity Bias removed . GLM echo gone.…
Freaky Frankenstein 5: Internal States. Beta Round 2: Final Testing. Fable 5 Uncensored- Positivity Bias removed . GLM echo gone. Anti-drafting updates. Dialogue overhauls.
- REDDIT · 2026-07-14 · 261 UPVOTES1.5x faster Gemma 4 NVFP4 Unsloth quants
1.5x faster Gemma 4 NVFP4 Unsloth quants
- REDDIT · 2026-02-10 · 252 UPVOTESGLM-4.7-Flash is now the #1 most downloaded model on Unsloth!
GLM-4.7-Flash is now the #1 most downloaded model on Unsloth!
- REDDIT · 2026-02-15 · 241 UPVOTESNanoGPT subscription changes (requests -> input tokens)
NanoGPT subscription changes (requests -> input tokens)
- REDDIT · 2026-10-01 · 206 UPVOTESPRESET WARS: Best preset right now? which one are you using?
PRESET WARS: Best preset right now? which one are you using?
- REDDIT · 2026-04-15 · 171 UPVOTESI need to vent about the available models and my RP journey. Feel free to ignore
I need to vent about the available models and my RP journey. Feel free to ignore
- REDDIT · 2026-04-09 · 152 UPVOTESI made Summaryception — a layered recursive memory system that fits 9,000+ turns into 16k tokens. It's free, it's open source, and it works…
I made Summaryception — a layered recursive memory system that fits 9,000+ turns into 16k tokens. It's free, it's open source, and it works with budget models.