Fast AI inference. 200+ models via one API. OpenAI/Anthropic compatible. 💬 Join Community → https://xitter.cf/t.co/5AzVHJpDHx 🚀 Get Higher Rate Limits → https://xitter.cf/t.co/5EvDyDEbaC

Singapore
Joined January 2024
🐰 Space Bunny just went open-weight! MiniMax-M3.1-Flash from @MiniMax_AI is now live on SiliconFlow — Day 0 of its open-weight release! 🚀 Stronger agents. Faster inference. Lower cost per task. ⚡ What's new? Agentic Coding: 73.7% on DeepSWE v1.1 Computer Use: 61.7 on OSWorld V2 Native Multimodal: Text, image & video inputs with 1M context Faster & More Efficient: Nearly 3× faster inference and 70% lower agentic task costs vs. M3 MSA Architecture: Sparse attention across all layers for greater efficiency More agent power. Less cost per task. Build with it → cloud.siliconflow.com/
2
222
💰 MiniMax-M3.1-Flash Pricing Per 1M tokens: Input: $0.15 Cached Input: $0.015 Output: $0.60
1
103
Same prompt, different models. DeepSeek V4.1 Flash 1017 vs. GPT-6 Astra — both building the same interactive 3D world with Three.js. Nearly the same quality, if not better — at over 50% lower cost than Astra. Prompt in the comments. 💬 Here's the side-by-side comparison 👇
1
3
539
Prompt 👇 Build a complete single-file Three.js/WebGL experience called “FLOATING WORLD”: a cinematic, interactive Eastern-fantasy island sanctuary floating above an endless cloud sea at golden hour. Use procedural geometry and textures only; no external 3D assets, images, paid APIs, placeholders, or TODOs. Three.js may load from a CDN. Handle missing WebGL, resizing, and mobile layouts gracefully. All visible text must be English only—no Chinese characters anywhere. SCENE Large central island with layered rocky underside, mossy edges, vines, terraced gardens, and distant floating islands. Central two-tier pavilion, arched Moon Bridge over a front-right pond, smaller grass pond to its left, ancient pine on the left, and two waterfalls at the island edge. Rich materials: weathered stone, dark timber, glazed dark-teal roof tiles, deep green moss/vegetation, water, restrained gold accents. Strong foreground/midground/background separation, atmospheric fog, soft shadows, warm rim light, controlled bloom. Entire island rig gently bobs up and down (±0.3 units, 11–18s periods). Cloud sea, distant islands, sky, and lighting remain stable. GRASS About 20,000 short grass clumps, each with around 8 curved blades. Clumps must overlap densely and read as a continuous, deep forest-green lawn, not pale spikes. Vary height, orientation, saturation, and lightness. Give the terrain itself a dark procedural grass-carpet texture so no brown soil is visible through gaps. Keep grass out of the pavilion, bridge, ponds, stream, and main path. PAVILION Two-tier Eastern pavilion with stone base, timber columns, railings, steps, altar, bell, warm lanterns, and two dark-red banners with gold borders. Both roofs must be dark glazed teal/green—never white or untextured. Add clear vertical tile channels from ridge to eave, strong tile texture, bump mapping, and actual raised 3D tile ribs. Keep the roof matte with low white specular. Add dark hip ridges, gold corner ornaments, and a gold finial. Banner monograms should be English “F” and “W,” not Chinese characters. PONDS AND WATER Main pond centered near (10, 8), radius ~5, with a raised stone rim and the arched stone Moon Bridge. Left grass pond near (5.5, 15.1), radius ~2, with a stone rim, reeds, stones, and grass around it. Both ponds and all shoreline stones must sit entirely inside the island silhouette. Remove any water/stone overhang and the small extra footbridge; shorten the stream. Fully cover both pond edges with stones, reeds, and grass—no exposed paper-like water edges. Water shader: procedural ripples, Fresnel reflection, sky reflection, sun highlights, subtle foam. Day: warm teal/green reflections. Twilight: deep indigo/blue-teal, never muddy yellow-green. Remove light-blue spotty shimmer. Night flow speed only ~4% of daytime. Day-to-twilight water must decelerate smoothly and monotonically, never speed up temporarily. Use an accumulated water-phase clock. WATERFALLS Left waterfall at ~70°: lower its top to about y=-2.6 and hide the start behind a large mossy cliff-rock curtain. Only the lower flow should be visible. Right waterfall at ~36°: connected to the bridge pond; keep its visible top unchanged. Use downward-flowing streaks, foam, soft edges, and mist at the base. CLOUDS Raise the cloud sea to about y=-20.4 so it wraps the island base. Build dense multi-layer white clouds: near-island bank, distant billows, low wisps, and mist. Clouds must look like overlapping cotton puffs, not flat paper layers; no visible holes in the middle. Front-left cloud cluster: ~18 puffs, partially covering the island, fluffy and volumetric. Four higher cloud groups above the island, each ~17–21 puffs, varied sizes and depths, tightly packed. All clouds white/warm-white. No gray cloud materials or gray shadow puffs. Do not restore the old right-side island-covering cloud. VEGETATION AND DETAILS Several dark-green pines/bonsai-like trees; one large Ancestor Pine with a crooked trunk, noble canopy, hanging lanterns, and a small stone shrine. Mossy shrubs, wildflowers, reeds, lilies, vines, warm floating motes, and a few distant birds. No animals other than the yellow dog below. YELLOW DOG Small golden-yellow dog: short chubby legs with rounded capsule shapes and smooth hip joints, short tail, rounded body/head, black nose, eyes, ears, tongue, red collar, gold tag. Animate running legs, body bob, tail wag, and head motion. Orient the dog from its actual movement direction; never sideways/backward. Give it a dedicated clear grass meadow centered near (-2.3, 9.2), ellipse approx. rx=3.0, rz=2.0. Remove the stone lantern near (-5.1, 10.5) and any stepping stones within a ~4.6 radius; replace removed stones with about 24 dark grass clumps. Dog randomly chooses targets inside this meadow, pauses, and chooses again—no fixed loop. It must never leave the meadow or intersect stones, buildings, ponds, or terraces. It rides the island bob. CAMERA AND INTERACTION Cinematic ~8-second opening reveal from a low side angle into the hero view. Hero camera ~(44, 31, 65), target ~(0, 2.8, 0), FOV 43°. OrbitControls with damping, drag-to-orbit, wheel zoom, no panning; add a smooth reset-camera button. Three clickable landmarks with smooth camera close-ups and English contextual cards: The Jade Pavilion: ~(18.5, 12.2, 17.2) → (0, 7.4, -2.6), FOV 40. The Moon Bridge: ~(25, 10, 32) → (10, 4.2, 8), FOV 42. The Ancestor Pine: ~(10, 15, 38) → (-10.5, 7, 5.4), FOV 42. Small hover label on landmarks; dragging must not trigger selection. LIGHTING AND UI Golden hour: warm sun from back-left, long shadows, peach horizon, cream clouds. Twilight: indigo sky, orange-pink horizon, dimmer sun, brighter lanterns, deep indigo water. Smooth 3.6-second lighting transition with an Atmosphere toggle. ACES tone mapping, restrained bloom, fog, soft shadows. English-only UI: “FLOATING WORLD,” “A Sanctuary Above the Clouds,” “The Realm of Drifting Clouds,” an F seal, atmosphere toggle, reset control, and three quiet-place links. Use an elegant serif/sans pairing with cream, ink, and gold. No Chinese text in UI, banners, labels, loader, or fallback. TECHNICAL Procedural CanvasTextures/shaders for stone, grass, wood, roof tiles, cloud puffs, water, waterfalls, banners. 60 FPS target on desktop, responsive, capped pixel ratio, working interactions, no runtime errors. Deliver a single polished HTML file.
241
🚀 DeepSeek-V4.1-Flash-1017 just landed on SiliconFlow! Stronger coding. Faster agent runs. Lower cost per task. ⚡ What's new? • Agentic Coding: 77.0% on DeepSWE v1.1, with stronger end-to-end coding capabilities. • Vibe Coding: Turn ideas into working projects, from games to 3D apps. • Faster Execution: Get tasks done in less time, with fewer output tokens. • Better Performance per Dollar: Pushing the performance–cost frontier further across agent benchmarks. Give it a spin👇 cloud.siliconflow.com
1
1,076
💸 DeepSeek-V4.1-Flash-1017 Pricing on SiliconFlow All prices per 1M tokens. Off-Peak (14:00–24:00 UTC) • Input: $0.15 • Output: $0.60 • Cached Input: $0.003 Peak (00:00–14:00 UTC) • Input: $0.30 • Output: $1.20 • Cached Input: $0.006
1
165
Not every model call needs an answer. Sometimes, it just needs a decision. 👏 Welcome Kev-4B to SiliconFlow. Kev-4B is the open-source community edition of Jev, built on Qwen3.5-4B for structured decision-making. Route. Rank. Approve. Escalate — without another generated response. No deployment or adaptation needed. One SiliconFlow API key, and Kev is ready to plug into your workflow. Huge shoutout to @jaredpalmer for open-sourcing Kev. ❤️ Try Kev-4B on SiliconFlow. ⚡️
1
6
837
The hottest open-source models on OpenRouter. Now live on SiliconFlow. 🔥 Meet Hy4 preview: 770B total parameters ⚡ 49B active per token 🧠 1M context 🛠️ Built for coding, analysis, research & complex real-world work 🔓 Apache 2.0 With SiliconFlow, plug Hy4 preview straight into the agents and tools you already use: Claude Code. Codex. Cursor — and more. Your tools. Hy4 preview. Powered by SiliconFlow. Fast, reliable inference — ready for real-world workloads.
3
1
8
1,138
Hy4 preview pricing on SiliconFlow 💰 Input: $0.834 / M tokens Cached input: $2.501 / M tokens Output: $0.042 / M tokens High-performance inference, without the flagship price tag. Start building with Hy4 preview on SiliconFlow.com
3
3
428
SiliconFlow Still Supports DeepSeek V4 Flash & V4 Pro — Keep Your Existing Workflows Running. DeepSeek’s V4 lineup is changing 👀 V4 Flash has been retired, with V4 Pro set to follow on Sep 14. Still using the original V4 models? No rush to say goodbye. V4 Pro 0813, V4 Pro (0424), and V4 Flash 0731 are all available on SiliconFlow. 🧵 With BYOK, your existing V4 workflows can keep running: Bring your SiliconFlow API key → pick the original V4 Flash / V4 Pro → keep building with the models already in your stack. Still need the original V4? You know where to find them :) siliconflow.com/models/deeps…
6
13
861
Replying to @deepseek_ai
@deepseek_ai just made Flash the new Pro. 🤯 DeepSeek-V4.1-Flash is live on SiliconFlow — Day 0. 552B MoE — ~8B active during prefill, ~16B during decode. ⚡ Faster inference, higher throughput 👁️ Native vision 🧠 1M context window 📉 ~4× smaller KV cache footprint vs. V4 Flash 📖 MIT licensed Run it on SiliconFlow — production-ready, high-throughput inference that keeps Flash fast at scale. Flash in name. Flagship in performance. Build with it → siliconflow.com
6
15
832
And the pricing is just as Flash. ⚡️ DeepSeek-V4.1-Flash on SiliconFlow: Off-peak · 14:00–24:00 UTC Cache → $0.003 / M tokens Input → $0.15 / M tokens Output → $0.60 / M tokens Peak · 00:00–14:00 UTC Cache → $0.006 / M tokens Input → $0.30 / M tokens Output → $1.20 / M tokens
152
DeepSeek V4 Flash just got vision. 👀 DeepSeek-V4-Flash-Vision-Exp is now live on SiliconFlow — and available via @OpenRouter too. Built on V4 Flash, now with visual understanding for multimodal tasks and agents. Strong text capabilities. Stronger multimodal agent performance. 1M context. MIT licensed. Build with DeepSeek Vision on SiliconFlow. 🚀
2
8
786
See it in action. 👀 Try DeepSeek-V4-Flash-Vision-Exp now on siliconflow.com.
1
351
10% cheaper. 20× more TPM. 🔥 Your limits on Kimi K3 just got boosted on SiliconFlow. Starting now: 💸 10% OFF Kimi K3 ⚡ Default TPM: 100K → 2M New pricing: Cache input $0.30/M → $0.27/M Input $3.00/M → $2.70/M Output $15.00/M → $13.50/M More capacity. Lower cost. Available to everyone! Build with Kimi K3 on siliconflow.com 🚀
4
1
14
1,136
💸$15.51 vs. $120.93. Same prompt. Same 3D game challenge. No manual code edits. GLM-5.3 completed the build at nearly 1/8 the cost of FABLE 5! Watch the side-by-side comparison—and decide for yourself. 👇 GLM-5.3 is now on SiliconFlow.
9
2
21
1,807
GLM-5.3 from @Zai_org is now open source — and live on SiliconFlow. 🚀 Day-0 support, as always. No extra scale. A lot more intelligence. 744B total / 40B active — the same base model as GLM-5.2, pushed much further through post-training. Competitive with Fable 5 and GPT-5.6 Sol across frontier coding and agentic benchmarks. SiliconFlow ranks #1 in GLM-5.2 token share on @OpenRouter. Now we’re bringing that experience to GLM-5.3. Build with GLM-5.3 now.
10
1
28
1,625
💰 GLM-5.3 pricing on SiliconFlow Cached input: $0.26 / M tokens Input: $1.40 / M tokens Output: $4.40 / M tokens
3
373