VOOZH about

URL: https://www.hardware-corner.net/gpu-llm-benchmarks/rtx-4080-super/

⇱ RTX 4080 SUPER Local LLM Benchmarks, Context Scaling & Supported Models 2026 – Hardware Corner

Tier 1 Enthusiast

RTX 4080 SUPER LLM Performance

Local LLM Performance: 52.8 t/s average on 14B models at 16k context.

Gen (14B 4-bit) 52.8 t/s
PP (14B 4-bit) 2,527 t/s
Max Model 20B
VRAM
16 GB GDDR6X
Bandwidth 736 GB/s
Token Gen (14B @ 4k Ctx)

52.8T/s

Prompt Proc (14B @ 4k Ctx)

2,527T/s

Summary

In our 2026 review, we view the RTX 4080 SUPER as a high-speed disappointment for local LLM workloads due to its pricing. While we measured good prompt processing speeds that exceed the RTX 3090 the 16GB VRAM limit makes absolutely no sense at the current price point. Although it generates tokens reasonably well, we believe it is a poor investment; for slightly more money, you can acquire a 24GB card that unlocks the next tier of model sizes (27B+), offering far superior longevity and versatility.

Key Insights

Capable of running gpt-oss 20B (MXFP4) with full 128k context support completely in VRAM.
Outperforms the RTX 3090 in prompt processing, achieving speeds over 2,500 t/s on 14B models.
Delivers solid token generation at 52.8 t/s (14B @ 16k context), though slightly slower than the RTX 3090.
Limited by 16GB VRAM, preventing the use of 27B+ parameter models that 24GB cards can handle.

Current Price in US

$930

Avg. Market Value

Current Pricing

Hardware Specs
VRAM 16GB GDDR6X
Capable of running 20B model
Bandwidth 736 GB/s
Architecture Ada Lovelace
Memory speed 23 Gbps
Memory bus 256 bit
TDP 320 W
Suggested PSU 750 W
Price/GB VRAM $58.13
Price/(t/s) with 14B @ 16k $17.63

Biggest LLMs You Can Run on This GPU

The models below represent the largest language models that fit fully in VRAM on this GPU using 4-bit quantization (GGUF). Benchmarks include token generation and prompt processing speeds measured at their maximum supported context length.

gpt-oss 20B (MXFP4) Max 128k
Token Generation 60.0 t/s @ 128k context
Prompt Processing 1,145.0 t/s @ 128k context
Qwen3 14B (Q4_K) Max 32k
Token Generation 42.6 t/s @ 32k context
Prompt Processing 1,769.3 t/s @ 32k context
Qwen3 8B (Q4_K) Max 64k
Token Generation 39.1 t/s @ 64k context
Prompt Processing 1,501.5 t/s @ 64k context

Note: Context values are grouped into standard tiers (4K, 16K, 32K, 64K, 128K). Models may support slightly higher context, but they remain in the lower tier unless they reach the next bracket.

RTX 4080 SUPER local LLM Inference Performance vs Similar GPUs

Compare prompt ingestion and token generation speeds against similar GPUs across widely used local models and extended context lengths up to 256K.

Local LLM Benchmarks

Prompt processing (t/s) and token generation speed (t/s) across different open weight models and context lengths.

Prompt Processing
Model 4k Ctx 16k Ctx 32k Ctx 64k Ctx 128k Ctx 256k Ctx
Qwen3 8B (Q4_K) 6,137.0 3,858.1 2,537.1 1,501.5 — —
Qwen3 14B (Q4_K) 3,745.0 2,526.7 1,769.3 — — —
gpt-oss 20B (MXFP4) 6,364.0 4,708.7 3,328.1 1,961.5 1,145.0 —
Token Generation
Model 4k Ctx 16k Ctx 32k Ctx 64k Ctx 128k Ctx 256k Ctx
Qwen3 8B (Q4_K) 104.2 79.4 59.5 39.1 — —
Qwen3 14B (Q4_K) 64.2 52.8 42.6 — — —
gpt-oss 20B (MXFP4) 139.1 123.0 107.2 81.5 60.0 —

Frequently Asked Questions

Common questions about running LLMs on the RTX 4080 SUPER.