Qwen3.5 9B Instruct (Q8)
Local OKOKB · 26
Best for: Quality, Coding, Reasoning·Pop: 86/100
Perf: est. 64 tok/s, typical error 5% · first token ~1.2s
Fits in 16 GB VRAM with room to spare. Best for quality, coding, reasoning on RTX 4070 Ti SUPER.
Best Local AI Models for RTX 4070 Ti SUPER (16GB)
The RTX 4070 Ti SUPER packs 16GB GDDR6X and delivers est. 94 tok/s, typical error 5% for 8B models. A strong performer from the previous generation, offering enough VRAM for 14B models with solid throughput.
Quick answer
The best local LLM for the RTX 4070 Ti SUPER is Qwen3.5 9B Instruct (Q8) (est. 64 tok/s, typical error 5%) on its 16GB VRAM. It uses ~10.7GB of VRAM; the RTX 4070 Ti SUPER handles up to 14B parameter models at Q4. A 14B model at Q4 runs at est. 58 tok/s, typical error 15%. (ModelFit, 2026)
Sizing rule: a Q4 model needs about 0.6 GB of VRAM per billion parameters, and ModelFit budgets 90% of this card's 16GB for weights, context, and KV-cache. The per-size table below uses that same budget. Other strong fits: Gemma 4 12B (12B, ~8GB) and Gemma 4 E4B (Q8) (4.5B, ~7.5GB). On this card Qwen3.5 9B Instruct (Q8) runs at est. 64 tok/s, typical error 5%. What it will not run: a 32B model does not fit in VRAM alone, so it spills to system RAM over PCIe and slows sharply.
$ollama run qwen3.5:9b-q8_0
- Top pick
- Qwen3.5 9B Instruct (Q8)
- Decode speed
- est. 64 tok/s, typical error 5%
- VRAM needed
- ~10.7 GB
Speeds are ModelFit estimates from memory bandwidth and model size, not measured benchmarks. Typical error: the median gap between the ModelFit speed model and measured runs on hardware the model was not fitted on, for this backend and model size.
Cite this page: ModelFit, RTX 4070 Ti SUPER 16GB Local LLM (2026): 14B at est. 58 tok/s, https://modelfit.io/gpu/rtx-4070-ti-super/, updated October 2026, CC BY 4.0.
Specs
VRAM
16 GB GDDR6X
Speed (8B Q4)
est. 94 tok/stypical error 5%
Bandwidth
672 GB/s
Architecture
Ada Lovelace
Max model size
Up to 14B parameter models
Compatibility
10 excellent, 0 workable
Table 1
RTX 4070 Ti SUPER estimated tokens/sec by model size
Q4_K_M · ModelFit estimate| Model size | Est. speed | Fit on 16GB |
|---|---|---|
| 7B | est. 107 tok/stypical error 5% | Fits in VRAM |
| 14B | est. 58 tok/stypical error 15% | Fits in VRAM |
| 20B MoE (3.6B active) | est. 125 tok/stypical error 15% | Fits in VRAM |
| 32B | est. 5 tok/stypical error 5% | CPU offload (slow) |
| 35B MoE (3B active) | est. 51 tok/stypical error 15% | CPU offload (slow) |
| 70B | est. 1 tok/stypical error 10% | CPU offload (slow) |
| 120B MoE (5.1B active) | est. 32 tok/stypical error 15% | CPU offload (slow) |
Fig. 1RTX 4070 Ti SUPER: estimated speed by model size
Decode speed, generic Q4_K_M model of each size at 4K context, 672 GB/s, typical error 5-15%
Fits in 16 GBCPU offload (slow)
7B107
14B58
20B MoE (3.6B active)125
32B5
35B MoE (3B active)51
70B1
120B MoE (5.1B active)32
tok/s est.ModelFit estimates, not measured benchmarks: decode speed of a generic Q4_K_M model of each size at 4K context on the RTX 4070 Ti SUPER's 672 GB/s bandwidth. MoE rows scale by active parameters (decode reads only the active experts), so a 35B MoE runs far faster than a dense 32B. "CPU offload" sizes exceed the 16GB VRAM; dense models slow to a crawl there, MoE models degrade less because hot experts stay GPU-resident. What typical error means.
Fig. 2
Context costs VRAM too.
Qwen3.5 9B Instruct (Q8) loads ~10.7 GB of weights; at 16k context the KV cache adds ~0.5 GB (still fits the ~14 GB usable VRAM), and at 64k it adds ~2.0 GB (still fits).
Weights 11 GBKV cache · 16k context 0.5 GBHeadroom 3.2 GBRuntime reserve 1.6 GB
The model leaves about 3.2 GB of the usable VRAM budget free after weights and 16k context.
KV-cache figures assume an fp16 cache, the llama.cpp/Ollama default. Standard GQA models use a size-class estimate (8 KV heads x 128 head dim class); hybrid linear-attention models (Qwen3.5/3.6, Qwen3-Next) use the exact per-token cost from their published config, since only their sparse full-attention layers cache KV. A q8_0 KV cache roughly halves either figure. Estimates, not measurements.
Get told when a model beats Qwen3.5 9B Instruct (Q8) on your RTX 4070 Ti SUPER
One email when a new open-weight model is a better fit for 16 GB of VRAM, plus the Thursday weekly.
You'll also get the weekly ModelFit email. By signing up you agree to our Privacy Policy. Unsubscribe from either email anytime, in one click.
Where to buy the RTX 4070 Ti SUPER
Storage & accessories for your model library
- Internal NVMe SSD · 2TBA Gen4 M.2 drive keeps your whole GGUF and quant collection on fast local storage, loading models straight off NVMe.Check price on Amazon
- USB4 NVMe Enclosure40Gbps external storage fast enough to run models from. Pair it with an M.2 drive for a portable model vault.Check price on Amazon
ModelFit may earn a commission on purchases through these links, at no extra cost to you.
RTX 4070 Ti SUPER VRAM for AI: what actually fits?
16GB GDDR6X at 672 GB/s positions this card between the 5060 Ti and 5070 Ti in bandwidth. It loads 14B Q4 models with room to spare and handles 8B models at est. 94 tok/s, typical error 5%. The main drawback is pricing: it launched at $799 MSRP, but 2026 memory-shortage pricing has pushed street prices well past that, above the newer RTX 5070 Ti, which is actually faster. Best bought used, where prices sit closer to the original MSRP.
Table 2
RTX 4070 Ti SUPER vs similar GPUs
| Hardware | Memory | Speed | Bandwidth |
|---|---|---|---|
| RTX 4070 SUPER | 12 GB | est. 73 tok/stypical error 5% | 504 GB/s |
| RTX 4070 Ti SUPER | 16 GB | est. 94 tok/stypical error 5% | 672 GB/s |
| RTX 4080 SUPER | 16 GB | est. 102 tok/stypical error 5% | 736 GB/s |
| RTX 5070 Ti | 16 GB | est. 125 tok/stypical error 5% | 896 GB/s |
Recommended models
registry-verified10 modelsGemma 4 12B
Local OKExcellentA · 44
Best for: Chat, Coding, Multimodal·Pop: 80/100
Perf: est. 67 tok/s, typical error 15% · first token ~0.3s
Fits in 16 GB VRAM with room to spare. Best for chat, coding, multimodal on RTX 4070 Ti SUPER.
Gemma 4 E4B (Q8)
Local OKExcellentA · 48
Best for: On-device, Mobile, Chat·Pop: 82/100
Perf: est. 101 tok/s, typical error 5% · first token ~1.0s
Fits in 16 GB VRAM with room to spare. Best for on-device, mobile, chat on RTX 4070 Ti SUPER.
Gemma 4 12B (Q8)
Local OKOKB · 11
Best for: Chat, Coding, Multimodal·Pop: 80/100
Perf: est. 42 tok/s, typical error 15% · first token ~1.6s
Fits in 16 GB VRAM with room to spare. Best for chat, coding, multimodal on RTX 4070 Ti SUPER.
Ministral 3 14B
Local OKExcellentA · 37
Best for: Chat, Reasoning, Vision·Pop: 74/100
Perf: est. 58 tok/s, typical error 15% · first token ~0.3s
Fits in 16 GB VRAM with room to spare. Best for chat, reasoning, vision on RTX 4070 Ti SUPER.
Qwen3.5 9B Instruct
Local OKExcellentA · 51
Best for: Quality, Coding, Reasoning·Pop: 86/100
Perf: est. 101 tok/s, typical error 5% · first token ~0.2s
Fits in 16 GB VRAM with room to spare. Best for quality, coding, reasoning on RTX 4070 Ti SUPER.
LFM2 24B-A2B Instruct
Local OKOKB · 3
Best for: Local AI agents, privacy-first tool calling, MCP workflows·Pop: 80/100
Perf: est. 194 tok/s, typical error 15% · first token ~0.1s
Fits in 16 GB VRAM with room to spare. Best for local ai agents, privacy-first tool calling, mcp workflows on RTX 4070 Ti SUPER.
Ornith 1.0 9B
Local OKExcellentA · 61
Best for: Agentic coding on small machines·Pop: 76/100
Perf: est. 101 tok/s, typical error 5% · first token ~0.2s
Fits in 16 GB VRAM with room to spare. Best for agentic coding on small machines on RTX 4070 Ti SUPER.
Qwen3 14B
Local OKOKB · 24
Best for: Coding, Quality·Pop: 84/100
Perf: est. 58 tok/s, typical error 15% · first token ~0.4s
Fits in 16 GB VRAM with room to spare. Best for coding, quality on RTX 4070 Ti SUPER.
LFM2.5 8B-A1B
Local OKExcellentA · 62
Best for: On-device agents, tool calling, multilingual chat·Pop: 72/100
Perf: est. 240 tok/s, typical error 15% · first token ~0.1s
Fits in 16 GB VRAM with room to spare. Best for on-device agents, tool calling, multilingual chat on RTX 4070 Ti SUPER.
Can I run it on the RTX 4070 Ti SUPER?
Per-model verdict pages for this exact card: weights, KV cache by context length, and estimated tok/s, each derived from the same engine the recommendations above use.
- Qwen3 8B Q4_K_Mfits
- Qwen3 14B Q4_K_Mfits
- GPT-OSS 20B MXFP4fits
- Qwen3.8 27B Q4_K_Mtight
- Llama 3.3 70B Instruct Q4_K_Mno
- Qwen3.5 35B-A3B Instruct Q4_K_Mtight
- Gemma 4 26B-A4B Q4_K_Mtight
- Qwen3.5 4B Instruct Q4_K_Mfits
- Gemma 4 E4B Q4_K_Mfits
- Gemma 3 4B Instruct Q4_K_Mfits
- Llama 3.1 8B Instruct Q4_K_Mfits
- Qwen3.5 9B Instruct Q4_K_Mfits
- Gemma 4 12B Q4_K_Mfits
- Mistral Nemo 12B Q4_K_Mfits
- Qwen3.5 9B Instruct (Q8) Q8_0fits
- Gemma 4 12B (Q8) Q8_0fits
- LFM2 24B-A2B Instruct Q4_K_Mfits
- Qwen3.5 27B Instruct Q4_K_Mtight
- Qwen3.6 27B Q4_K_Mtight
- Gemma 4 31B Q4_K_Mtight
- Qwen3.6 35B-A3B Q4_K_Mno
- Qwen3.8 27B (Q8) Q8_0no
Similar GPUs for local AI
- RTX 5070 Ti16GB · est. 125 tok/s
- RTX 4080 SUPER16GB · est. 102 tok/s
- RTX 4070 SUPER12GB · est. 73 tok/s
Speeds are estimates for Qwen3 8B Q4_K_M, typical error 5%.
Compatible model families
- QwenAlibaba Cloud: Qwen3.8 27B local flagship (Apache 2.0, vision)
- LlamaMeta: Most popular open-weight model family
- DeepSeekDeepSeek AI: V4.1 Flash: multimodal 552B MoE, API only
- MistralMistral AI: Excellent performance-per-parameter ratio
- GemmaGoogle DeepMind: Gemma 4 E4B: 4.5B effective, runs on 8GB Macs
- PhiMicrosoft: Best quality-per-gigabyte at small sizes
RTX 4070 Ti SUPER FAQ: common questions
How much VRAM does the RTX 4070 Ti SUPER have for LLMs?
The RTX 4070 Ti SUPER has 16GB GDDR6X VRAM with 672 GB/s bandwidth. About 15.5GB usable for model loading. Fits 14B models at Q4 and all 7B-9B models comfortably.
What size LLM can I run on an RTX 4070 Ti SUPER?
Up to 14B parameter models at Q4 quantization. Same model capacity as other 16GB cards. Speed on 8B models is est. 94 tok/s, typical error 5% (14B around est. 58 tok/s, typical error 15%), faster than the 4060 Ti but slower than the newer 5070 Ti.
Is the RTX 4070 Ti SUPER still worth buying for AI in 2026?
At current street pricing, well above its $799 launch MSRP due to the 2026 memory shortage, no. The RTX 5070 Ti costs less and is 33% faster. However, a used 4070 Ti SUPER can still be good value: you get 16GB VRAM and est. 94 tok/s, typical error 5% on 8B models at a reasonable price.
RTX 4070 Ti SUPER vs RTX 5070 Ti for local AI?
The RTX 5070 Ti is faster (est. 125 tok/s, typical error 5% vs est. 94 tok/s on 8B), cheaper, and uses GDDR7. The 5070 Ti wins on every metric for AI workloads. The only reason to buy the 4070 Ti SUPER is availability or used pricing.
How fast is a 27B-class model on the RTX 4070 Ti SUPER?
The RTX 4070 Ti SUPER's 16GB of VRAM cannot fit a 32B model comfortably. The largest size class it fits is 20B MoE (3.6B active), at est. 125 tok/s, typical error 15%. The largest individual model in the catalog that fits is LFM2 24B-A2B Instruct (24B).
Related guides & benchmarks
Sizing local AI? Start with RAM & VRAM
Browse all NVIDIA GPUs for AI
Real-world calibration
Have this hardware? Share a measured run.
Measure a fixed Ollama reference model on your machine. This separate, opt-in contribution does not change ModelFit's third-party SWE-Bench leaderboard; results appear on the measured-speed leaderboard after the next site update.
$ npx -y @wecko-ai/modelfit bench --submit --cleanup
Want personalized recommendations?
Use our interactive wizard to compare models across Apple Silicon and NVIDIA GPUs.